Coupled and decoupled fronto–cerebellar and affective brain networks characterize adolescents with borderline personality disorder

Coupled and decoupled fronto–cerebellar and affective brain networks characterize adolescents with borderline personality disorder | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Coupled and decoupled fronto–cerebellar and affective brain networks characterize adolescents with borderline personality disorder Xiaoping Yi, Xue Liu, Xueying Wang, Jinfan Zhang, Feifei Wu, Liying Shen, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8860480/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Introduction Previous work demonstrated that adolescents with borderline personality disorder (aBPD) exhibit gray and white matter density (GM) alterations in multiple regions, yet prior neuroimaging studies have predominantly examined GM and WM in isolation, limiting the ability to capture their joint neurodevelopmental alterations. The present study aims to overcome this limitation by integrating GM and WM structural features within a unified unsupervised multimodal data-fusion framework, in order to identify latent brain components reflecting shared and dissociable patterns and to relate these multimodal signatures to clinical and functional measures. We predicted that adolescents with BPD would show coupled and decoupled gray-white matter networks related to affective processing and executive functions. Methods High-resolution T1-weighted structural MRI data from 129 adolescents with borderline personality disorder (aged 12–17) and 107 age-, gender-, and education-matched healthy controls were analyzed using transposed independent vector analysis (tIVA-G), integrating gray and white matter features. Multiple clinical measures were collected and used to assess the clinical relevance of the identified multimodal components. In addition, decision tree models were applied to derive an interpretable predictive model discriminating adolescents with BPD from healthy controls. Results Two distinct structural components differentiating adolescents with BPD from healthy controls were identified. A first component (IC18) reflected a gray matter-driven pattern, characterized by reduced volume in a fronto-cerebellar loop, with no corresponding white matter contribution. A second component (IC5) captured a coupled gray matter–white matter alteration, showing increased gray matter in limbic–affective regions and Default network regions, including temporal and medial frontal areas, precuneus, and cerebellum, alongside reduced white matter in frontal regions. This network was significantly associated with greater emotion dysregulation, self-injurious behaviors, internalizing symptoms, and lower global functioning. Decision tree discriminated adolescents with BPD from healthy controls with moderate accuracy (71.7%). Conclusions Together, these multimodal patterns link large-scale brain organization to emotion dysregulation, self-injurious behaviors, and functional impairment in youth with BPD, providing a more comprehensive neurobiological account of the disorder and informing future biomarker-oriented and mechanistically grounded interventions. Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Borderline personality disorder (BPD) represents a highly prevalent and clinically burdensome condition, affecting approximately 1–6% of the general population, with substantially higher rates reported in outpatient (up to 10%) and inpatient psychiatric settings (up to 25%) (Ellison et al., 2018 ). Increasing attention has been devoted to borderline personality disorder in adolescence (aBPD), which is now recognized as a valid and clinically meaningful condition rather than a transient developmental phase. aBPD is primarily characterized by pervasive difficulties in emotion regulation and a high incidence of non-suicidal self-injury (NSSI) (Bohus et al., 2021 ; Mirkovic et al., 2021 ). Emotion regulation difficulties in aBPD encompass a broad range of deficits, including impaired emotional awareness and understanding, heightened emotional reactivity, reduced capacity to modulate affective responses, and limited access to adaptive regulatory strategies. Clinical observations indicate that adolescents with BPD frequently exhibit several, if not all, of these impairments. Importantly, converging theoretical and empirical evidence suggests that such emotion regulation failures may play a causal role in the emergence and maintenance of NSSI behaviors (Dadomo et al., 2018 ). Despite the clinical relevance of aBPD, its neurobiological underpinnings, particularly the neural correlates of emotion dysregulation and NSSI, remain insufficiently characterized. This gap is largely attributable to the fact that the majority of neuroimaging studies on BPD have been conducted in adult populations. Structural MRI investigations in adults with BPD have consistently reported gray matter alterations within limbic–frontal circuits, including reduced volumes in the bilateral hippocampus, dorsolateral prefrontal cortex, and medial frontal cortex (Schulze et al., 2016 ; Rossi et al., 2013 ; Prossin et al., 2010; Minzenberg et al., 2008; Xu et al., 2016 ; Soloff et al., 2008 ; Luo et al., 2010; Ding et al., 2021; Grecucci et al., 2021; 2023 ; 2024). These structural abnormalities are functionally meaningful, as they map onto core clinical features of BPD, such as affective instability, impaired impulse control, and disturbed interpersonal functioning (Schulze et al., 2016 ; Dadomo et al., 2016 , 2018 ; Frederickson et al., 2018 ; Ruocco et al., 2013 ). In addition to limbic and posterior alterations, functional neuroimaging studies have implicated abnormal activity within the default mode network (DMN), particularly increased resting-state engagement, in affective dysregulation and maladaptive self-referential and interpersonal processes in BPD (Ruocco et al., 2013 ; Paris et al., 2007 ; Koenigsberg et al., 1999 ). However, despite growing recognition of adolescence as a critical neurodevelopmental window for the emergence of BPD-related pathology, only a limited number of studies have specifically examined the neural architecture of aBPD. Adolescence represents a critical neurodevelopmental period characterized by profound structural and functional brain reorganization. Investigating borderline personality disorder during this developmental window is therefore essential to improve early detection and to inform timely and targeted interventions. Neuroimaging studies focusing on adolescents with BPD have provided initial evidence of gray matter abnormalities within limbic and frontal circuits, including alterations in the bilateral orbitofrontal and dorsolateral prefrontal cortices, right amygdala, left anterior cingulate cortex, and bilateral hippocampus (Chanen et al., 2008 ; Richter et al., 2014 ; Brunner et al., 2010 ; Goodman et al., 2011 ; Yi et al., 2023 ). Nevertheless, the involvement of large-scale networks such as the default mode network (DMN) and the precise contribution of prefrontal control regions in aBPD remain insufficiently established. Beyond cortico-limbic regions, a growing body of evidence has highlighted the cerebellum as a key node in affective and cognitive regulation across a range of psychiatric conditions (Salvador et al., 2011 ; Koziol et al., 2014 ; Hallett et al., 2016 ). Importantly, cerebellar contributions to emotion and cognition are thought to be implemented through recurrent frontal–cerebellar loops linking medial and lateral prefrontal regions with posterior cerebellar territories, supporting predictive control, error monitoring, and adaptive regulation of internal states (e.g., Schmahmann, 2019; Stoodley & Schmahmann, 2010 ). Within this framework, the cerebellum is not conceived as an isolated affective structure, but as a modulatory node embedded in fronto–cerebellar loops that refine prefrontal control over emotional and self-referential processes (Schmahmann, 2019). Disruption of these loops has been proposed to contribute to psychopathological phenotypes characterized by impaired emotion regulation, impulsivity, and unstable self-representation. In adult BPD, cerebellar alterations have been associated with emotional dysregulation and impulsivity (Schulze et al., 2016 ; Grecucci et al., 2022 ; 2025), raising the question of whether similar cerebellar involvement is already detectable during adolescence, when core symptoms first emerge. aBPD may involve early alterations in fronto–cerebellar systems supporting the integration of emotional monitoring and cognitive control (Yi et al., 2026). Despite these advances, the existing neuroimaging literature on aBPD is characterized by several methodological shortcomings. A primary limitation is the widespread reliance on mass-univariate analytical approaches, which treat each voxel as statistically independent and therefore fail to capture the coordinated nature of brain organization (Sorella et al., 2019 ; Lapomarda et al., 2021 , 2022; Grecucci et al., 2022 , 2023 ). Most studies have employed voxel-based morphometry (VBM), a technique that may be insufficiently sensitive to subtle but spatially distributed patterns of structural alteration (Pappaianni et al., 2018 ). Furthermore, the generalizability of previous findings has been limited, reducing their potential translational value. Additional constraints include relatively small sample sizes (typically ranging from 13 to 53 participants), a strong overrepresentation of female patients (often exceeding 75–100%), and incomplete control of key sociodemographic variables such as educational level. Finally, the restricted assessment of clinical and psychological dimensions in several studies has hampered a comprehensive interpretation of observed neural abnormalities. Together, these limitations underscore the need for large-sample, multivariate, and clinically well-characterized investigations capable of capturing distributed neurodevelopmental alterations in aBPD. In a recent large-scale neuroimaging study, Yi and colleagues ( 2025 ) investigated structural brain alterations in adolescents with borderline personality disorder using a multivariate, machine-learning–based framework by combining source-based morphometry with supervised learning techniques. Their results revealed a distinctive pattern of gray matter covariance, characterized by increased volume within regions overlapping the posterior hub of the default mode network and the cerebellum, alongside reduced gray matter in frontal control regions. For what concerns white matter alterations, previous work in adolescents with borderline personality disorder reported reduced fractional anisotropy and increased mean diffusivity within cortical–limbic and cortical–thalamic pathways compared to healthy controls (Yi et al., 2024 ), by using Diffusion tensor imaging, suggesting early alterations in WM organization. Disruptions in these WM properties have been proposed to impair large-scale network connectivity, with downstream consequences for emotion regulation and impulse control in BPD (Grottaroli et al., 2020; Kelleher-Unger et al., 2021). Consistent with this view, DTI studies and meta-analytic evidence in adults with BPD have identified WM abnormalities predominantly affecting the corpus callosum, corona radiata, external capsule, and uncinate fasciculus (Grottaroli et al., 2020; Kelleher-Unger et al., 2021). Increased RD within the corpus callosum has been linked to anxiety severity (Quattrini et al., 2019), while alterations in RD and AD have been reported across thalamic radiations, superior longitudinal fasciculus, and inferior fronto-occipital fasciculus (Gan et al., 2016; Maier-Hein et al., 2014; Ninomiya et al., 2018). In a recent diffusion MRI study, Yi and colleagues ( 2025 ) investigated white matter microstructural alterations associated with emotional dysfunction and childhood trauma in adolescents with borderline personality disorder. Using tract-based spatial statistics on diffusion tensor imaging data, the authors reported a pattern of altered radial and axial diffusivity within key cortico–limbic pathways, including the corpus callosum, corona radiata, external capsule, and uncinate fasciculus. Specifically, adolescents with BPD showed increased radial diffusivity and reduced axial diffusivity in several callosal and fronto-limbic tracts compared to healthy controls. These findings provide evidence for early white matter abnormalities in adolescent BPD and suggest that WM microstructural measures may represent promising neurobiological markers linked to both clinical severity and adverse developmental experiences. One additional limitation of the previous study, is that they typically examined gray and white matter alterations separately, preventing a unified characterization of their joint contribution to emotion dysregulation and other symptoms. Converging evidence from adults with BPD points to a direction in which both structural properties of the brain are altered in these populations (Grecucci et al. 2023 ). As far as we know this methodology was never applied to adolescents with BPD. Building directly on our prior findings, the present study was therefore designed to extend this framework by integrating gray and white matter structural information within a unified unsupervised data-fusion approach, with the aim of identifying latent multimodal neurodevelopmental patterns underlying adolescent BPD. In recent years, machine learning (ML) has become increasingly central to neuroimaging data analysis, influencing not only methodological practice but also conceptual models of brain organization and disease (Nenning & Langs, 2022; Baggio et al., 2023 ). Group-level inference methods are often poorly suited to capturing individual variability and distributed network-level alterations, both of which are essential for understanding complex cognitive and behavioral phenomena (Grecucci et al., 2022 ; Hebart & Baker, 2018 ; Vieira et al., 2020 ; Schnack, 2020 ). Within this context, unsupervised learning techniques offer a powerful means of uncovering latent data structure through clustering and dimensionality reduction without relying on predefined outcome variables (Kherif & Latypova, 2020 ). To test the presence of coupled and decoupled GM-WM networks in aBPD, we applied an unsupervised machine-learning data-fusion approach based on Transposed Independent Vector Analysis (tIVA) to identify independent neural circuits across modalities (Adali et al., 2015; Lee et al., 2008 ). tIVA was selected because it integrates key advantages of independent component analysis and canonical correlation analysis while overcoming central limitations of existing fusion frameworks. Unlike joint independent component analysis, which imposes a single mixing matrix across modalities and assumes equivalent signal contributions (Calhoun et al., 2006 ), tIVA estimates modality-specific mixing matrices while statistically coupling source vectors, allowing shared variance to emerge without enforcing identical spatial patterns. Compared with multiset-CCA or mCCA+jICA approaches (Sui et al., 2013), tIVA incorporates higher-order, non-Gaussian statistics, increasing sensitivity to subtle and potentially nonlinear cross-modal relationships. Benchmark studies have demonstrated that tIVA yields more stable components and improved diagnostic classification relative to joint-ICA or linked-ICA, while maintaining computational scalability as additional modalities are included (Adali et al., 2015). tIVA provides a principled framework for capturing distributed GM–WM covariance patterns that are more functionally informative than traditional atlas-based representations (Biswal et al., 2010; Fox et al., 2005 ; Baggio et al., 2023 ; Grecucci et al., 2022 ). We expected to identify increased GM-WM concentration in regions belonging to the default mode network and to the cerebellum, together with reduced prefrontal control areas, consistent with prior functional and structural evidence in BPD (Koenigsberg et al., 1999 ; Paris et al., 2007 ; Ruocco et al., 2013 ; Schulze et al., 2016 ). We expected that alterations would relate to specific facets of emotion regulation and executive functioning. We further hypothesized the presence of a multimodal gray matter–white matter component reflecting coupled alterations within limbic–affective circuits, alongside complementary changes in frontal and posterior default mode regions, in line with previous reports of prefrontal and white matter involvement in BPD (Hazlett et al., 2005 ; Goethals et al., 2005 ; Schulze et al., 2016 ; Grecucci et al., 2023 ). Finally, we predicted that the expression of these data-driven components would be meaningfully associated with core clinical features of adolescent BPD, particularly emotion dysregulation and self-injurious behaviors, and would contribute to distinguishing adolescents with BPD from healthy controls. Methods Participants Adolescents with borderline personality disorder (BPD) were consecutively recruited between October 2021 and October 2023 from the psychiatric clinics of the Mental Health Center at Xiangya Hospital, Central South University. Healthy control participants (HCs), matched for age, gender, and education level, were recruited from local schools during the same period.Inclusion criteria for the BPD group were: (1) age between 13 and 18 years; (2) fulfillment of DSM-IV diagnostic criteria for BPD (≥ 5 out of 9 criteria); (3) symptom stability for at least two years; and (4) a score ≥ 66 on the Borderline Personality Feature Scale for Children (BPFS-C). Diagnosis followed a rigorous two-step procedure. First, an associate professor of psychiatry with over 15 years of clinical experience applied DSM-5-TR criteria during routine clinical assessment. All participants meeting initial criteria subsequently completed the DSM-IV Axis II Structured Clinical Interview for Personality Disorders (SCID-II), administered by trained child and adolescent psychiatrists with at least five years of clinical experience. Final diagnoses were independently confirmed by two senior psychiatrists, ensuring adherence to internationally recognized diagnostic standards. To enhance internal validity in structural MRI analyses, exclusion criteria were applied for psychiatric and neurodevelopmental conditions known to produce strong, independent neuroanatomical alterations, including schizophrenia spectrum disorders, bipolar spectrum disorders, major depressive disorder, post-traumatic stress disorder, attention-deficit/hyperactivity disorder, substance use disorders, and neurological diseases. These exclusions were implemented to reduce neurobiological confounds rather than to imply a comorbidity-free presentation of adolescent BPD. Healthy controls underwent comprehensive screening, including medical history review and psychological assessment, to exclude any current or past psychiatric or neurological disorders. Additional exclusion criteria for all participants included IQ ≤ 80, contraindications to MRI, substance use, and left-handedness, as assessed by the Edinburgh Handedness Inventory. The final sample comprised 129 adolescents with BPD (aged 12–17) and 107 matched healthy controls. A priori power analysis indicated that this sample size provides 80% power to detect medium effect sizes (Cohen’s d ≈ 0.5) at α = .05. Patients were recruited from a large tertiary psychiatric center receiving referrals from outpatient clinics, emergency services, community mental health providers, and inpatient units, ensuring a clinically representative sample of moderate-to-severe adolescent BPD. The study was approved by the Institutional Review Board of Xiangya Hospital, Central South University (IRB No. 2022020227). Written informed consent was obtained from the parents or legal guardians of all participants. Clinical assessment Diagnostic assessment was conducted using standardized, structured interviews. The borderline personality disorder diagnosis was established using the BPD module of the Structured Clinical Interview for DSM-IV Personality Disorders (SCID-II), with diagnosis confirmed when at least five of the nine criteria were met. Axis I psychiatric disorders were assessed using the Schedule for Affective Disorders and Schizophrenia for School-Age Children—Present and Lifetime Version (K-SADS-PL). Healthy control participants underwent the same structured interviews to exclude any personality or Axis I disorders. Interviews were conducted with both adolescents and their parents or legal guardians. Final diagnoses were determined following a comprehensive clinical evaluation by two experienced child and adolescent psychiatrists (QX and FJ, with 10 and 8 years of clinical experience, respectively). Information regarding psychiatric disorders in first-degree relatives was collected from participants and their parents or legal guardians. All clinical assessments were completed prior to MRI acquisition and on the same day. Intellectual functioning was evaluated using the Abbreviated Wechsler Intelligence Scale (Wechsler, 1999 ). Borderline personality traits were assessed using the Borderline Personality Features Scale for Children (BPFS-C), with a cutoff score of 66 indicating clinically significant borderline features. The BPFS-C includes four subscales assessing affective instability, identity problems, negative relationships, and self-harm. Emotion regulation difficulties were measured using the Difficulties in Emotion Regulation Scale (DERS; Li et al., 2018 ), which comprises six subscales covering emotional awareness, emotional clarity, nonacceptance of emotional responses, impulse control difficulties, difficulties in goal-directed behavior, and limited access to emotion regulation strategies. Anxiety symptoms were evaluated using the Screen for Child Anxiety Related Emotional Disorders (SCARED; Birmaher et al., 1999 ), which includes five subscales assessing panic/somatic symptoms, generalized anxiety, separation anxiety, social phobia, and school phobia. Global functioning was assessed with the clinician-rated Global Assessment of Functioning (GAF) scale, providing a composite index of psychological, social, and occupational functioning (Aas, 2011 ). Depressive symptoms were measured using the Mood and Feelings Questionnaire (MFQ; Wood et al., 1995 ), a 33-item self-report scale with higher scores indicating greater depressive severity. MRI data acquisition Structural MRI data were acquired on a Siemens MAGNETOM Prisma 3T scanner. For each participant, a high-resolution three-dimensional T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) sequence with whole-brain coverage was collected. Acquisition parameters were as follows: acquisition matrix 256 × 256 mm², isotropic voxel size 10 mm³, repetition time (TR) = 2300 ms, echo time (TE) = 203 ms, flip angle = 9°, and 176 contiguous sagittal slices. All images were visually inspected by a senior neuroradiologist (ZH; >30 years of experience) to ensure adequate image quality and to exclude incidental brain abnormalities. Preprocessing After initial quality control to exclude images with artifacts, all MRI data were preprocessed using a unified pipeline implemented in the Computational Anatomy Toolbox (CAT12; http://www.neuro.uni-jena.de/cat/ ) running within SPM12 ( http://www.fil.ion.ucl.ac.uk/spm/ ) and MATLAB. Images were segmented into gray matter, white matter, and cerebrospinal fluid, and modulated normalized maps were generated. High-dimensional spatial registration was performed using Diffeomorphic Anatomical Registration through Exponential Lie Algebra (DARTEL), followed by normalization to Montreal Neurological Institute (MNI) space. Finally, the normalized images were spatially smoothed using an 8-mm full-width at half-maximum Gaussian kernel. Transposed Independent Vector Analysis To identify distributed neural circuits underlying adolescent BPD, we employed an unsupervised machine-learning approach based on Transposed Independent Vector Analysis (tIVA) (Adali et al., 2015; Lee et al., 2008 ). tIVA is a multivariate source separation technique specifically designed for multimodal data fusion, allowing the joint decomposition of multiple imaging modalities—here gray matter (GM) and white matter (WM)—into statistically independent components while preserving modality-specific information. In contrast to conventional univariate or concatenation-based approaches, tIVA estimates separate mixing matrices for each modality while statistically coupling the corresponding source vectors, thereby enabling the identification of latent GM–WM covariance patterns without imposing identical spatial distributions across modalities. The tIVA analysis was implemented using the Fusion ICA Toolbox (FIT; http://mialab.mrn.org/software/fit ) (Calhoun et al., 2006 ) within the MATLAB 2018a environment (The MathWorks, Natick, MA, USA). Preprocessed GM and WM maps were entered simultaneously into the model. Prior to decomposition, the optimal number of independent vectors was estimated using the minimum description length (MDL) criterion, which provides a data-driven balance between model complexity and explanatory power. Independent Vector Analysis (IVA) extends Independent Component Analysis (ICA) to the joint analysis of multiple related datasets by explicitly modeling statistical dependence across datasets, in addition to enforcing independence within each dataset (Adali et al., 2015; Lee et al., 2008 ). In this framework, each dataset \(\:{X}^{\left(k\right)}\) (with \(\:k=1,\dots\:,K\) ) is assumed to arise from a linear mixture of latent sources \(\:{S}^{\left(k\right)}\) through modality-specific mixing matrices \(\:{A}^{\left(k\right)}\) . Source estimates are obtained via corresponding demixing matrices \(\:{W}^{\left(k\right)}\) , yielding subject- or voxel-wise source estimates \(\:{u}^{\left(k\right)}\) . A defining feature of IVA is the introduction of the source component vector (SCV), which is constructed by grouping corresponding sources across the Kdatasets. Each SCV therefore represents a multivariate entity that captures statistical dependence across modalities for a given component, while different SCVs are assumed to be mutually independent. This formulation allows IVA to exploit cross-dataset correlations when they are present, while reducing to standard ICA when a single dataset is analyzed (K=1). The IVA objective function is defined in terms of the mutual information rate across SCVs. Minimization of this criterion simultaneously promotes independence between different SCVs and dependence within the elements of each SCV. As a result, IVA can leverage multiple forms of statistical diversity, including higher-order statistics and sample dependence, thereby improving source identifiability and separation performance relative to ICA. Importantly, this property allows IVA to identify sources that may be Gaussian within individual datasets but become identifiable due to correlations across datasets—an advantage that is particularly relevant in multimodal fusion settings. Transposed Independent Vector Analysis (tIVA) further generalizes this framework by applying IVA to transposed data representations, such that the independent sources correspond to subject-wise expression profiles rather than spatial maps. In this formulation, tIVA effectively extends multi-set canonical correlation analysis (MCCA) by relaxing orthogonality constraints and incorporating higher-order statistical information. The tIVA output consists of subject-specific loading coefficients and spatial maps for each modality. Spatial maps were transformed into Talairach space to facilitate anatomical labeling and interpretation of the associated brain regions. Both positive and negative voxel weights were retained, reflecting regions contributing in opposite directions within each component. For visualization, component maps were rendered using Surf Ice ( https://www.nitrc.org/projects/surfice/ ) (Rorden), allowing intuitive inspection of the spatial extent and anatomical distribution of the identified networks. Group-level statistical analyses were performed on the component loading coefficients using independent-samples t-tests implemented in JASP (version 0.16.3; JASP Team, 2023) to assess differences between adolescents with BPD and healthy controls. This analytic strategy enabled us to quantify how specific multimodal network components differentiated diagnostic groups and to characterize their contribution to the structural phenotype of adolescent BPD. Predictive model Subject-specific loading coefficients derived from tIVA decomposition were subsequently entered into a supervised machine learning framework to classify adolescents with borderline personality disorder (aBPD) versus healthy controls (HCs). Supervised classification was implemented using a decision tree classifier as provided in JASP [65]. In this framework, diagnostic group membership (aBPD vs HC) was modeled as a function of the gray and white matter networks loadings. The decision tree algorithm recursively partitions the predictor space by selecting, at each step, the network component and corresponding threshold that best separates participants into subgroups with more homogeneous class labels. For each candidate split, the algorithm evaluates classification impurity—defined as the degree of class mixing within each node—using standard criteria such as misclassification error or Gini impurity. The split that maximally reduces impurity, that is, that produces the greatest improvement in class separation between aBPD and HC participants, is selected. As a result, each terminal node of the tree represents a subgroup of participants characterized by a specific pattern of network expression and a predominant diagnostic label. The predicted class for a new participant corresponds to the majority class within the terminal node to which that participant is assigned. Model generalizability was assessed using a hold-out validation strategy, with 80% of the observations used to train the classifier and the remaining 20% reserved for testing its performance on previously unseen cases. This procedure ensured that classification accuracy reflected true out-of-sample performance. Model hyperparameters were set as follows: a minimum of 20 observations was required to attempt a split, terminal nodes were constrained to contain at least 7 observations, maximum tree depth was set to 30, and the complexity parameter was fixed at 0.01 to penalize overly complex trees. All predictors were standardized prior to model estimation. Results Behavioral findings Demographic and clinical characteristics of adolescents with borderline personality disorder and healthy controls are reported in Table 1 . The two groups did not differ significantly in age (Mann–Whitney U = 7621, p = 0.154), years of education (U = 6200, p = 0.163), or gender distribution (χ² = 3.217, p = 0.073). In contrast, adolescents with BPD showed markedly elevated symptom severity across all clinical measures, including borderline personality features (BPFS), depressive symptoms (MFQ), anxiety symptoms (SCARED), and difficulties in emotion regulation (DERS), alongside significantly lower levels of global functioning as indexed by GAF scores, relative to healthy controls (all p < 0.001). Behavioral and clinical findings Adolescents with borderline personality disorder (aBPD) and healthy controls (HCs) did not differ significantly in basic demographic variables, including gender distribution (aBPD: 42 males, 87 females; HC: 47 males, 60 females; p = 0.073), age (mean age: aBPD = 15.55 years; HC = 15.90 years; p = 0.154), or years of education (aBPD = 9.82 years; HC = 9.65 years; p = 0.163). The mean duration of illness in the aBPD group was 12.85 years.In contrast, pronounced differences emerged across all clinical and psychological measures. Adolescents with aBPD showed markedly elevated borderline personality features, as indexed by the Borderline Personality Features Scale for Children (BPFS-C). Specifically, the total BPFS-C score was substantially higher in the aBPD group compared to HCs (86.67 vs. 56.00, p < 0.001). This pattern was consistent across all BPFS-C subscales, including affective instability (BPFS-C-1: 23.50 vs. 14.16, p < 0.001), identity problems (BPFS-C-2: 21.84 vs. 14.71, p < 0.001), negative relationships (BPFS-C-3: 20.76 vs. 14.29, p < 0.001), and self-harm (BPFS-C-4: 20.75 vs. 12.82, p < 0.001).Adolescents with aBPD also exhibited significantly greater difficulties in emotion regulation, as measured by the Difficulties in Emotion Regulation Scale (DERS). The total DERS score was markedly higher in the aBPD group compared to HCs (123.90 vs. 80.32, p < 0.001). All DERS subdomains showed significant group differences, including emotional awareness (DERS-A: 19.41 vs. 15.74, p < 0.001), emotional clarity (DERS-B: 15.32 vs. 10.41, p < 0.001), nonacceptance of emotional responses (DERS-C: 18.38 vs. 11.95, p < 0.001), impulse control difficulties (DERS-D: 21.20 vs. 11.49, p < 0.001), difficulties engaging in goal-directed behavior (DERS-E: 20.62 vs. 12.68, p < 0.001), and limited access to emotion regulation strategies (DERS-F: 29.89 vs. 17.04, p < 0.001).Anxiety-related symptoms were also significantly elevated in adolescents with aBPD, as assessed by the Screen for Child Anxiety Related Emotional Disorders (SCARED). The total SCARED score was more than doubled in the aBPD group relative to HCs (46.35 vs. 19.16, p < 0.001). This difference was observed across all examined anxiety domains, including panic/somatic symptoms (SCARED-A: 15.00 vs. 4.48, p < 0.001), generalized anxiety (SCARED-B: 11.51 vs. 4.98, p < 0.001), and separation anxiety (SCARED-C: 6.50 vs. 2.83, p < 0.001). Overall, these results indicate that adolescents with aBPD present a pervasive and multidimensional clinical profile characterized by severe borderline personality features, marked emotion regulation deficits, and elevated anxiety symptomatology, despite a demographic profile comparable to that of healthy peers. Neural findings Application of transposed Independent Vector Analysis (tIVA) to gray and white matter structural MRI data identified 20 independent vectors (tIVs) reflecting covarying neuroanatomical patterns across subjects. Group differences between adolescents with borderline personality disorder (aBPD) and healthy controls (HCs) were assessed on the subject-specific loading coefficients using independent-samples t-tests. To account for multiple comparisons, a Bonferroni-corrected significance threshold was applied. At this corrected threshold, two components showed significant group differences, corresponding to a gray matter–only frontal–cerebellar network (tIV18) and a coupled gray–white matter affective network (tIV05). tIV18 showed a robust group difference in gray matter loadings (t = 4.131, df = 231, p < 0.001). In contrast, no significant group difference was observed for the white matter loadings of this component (t = 0.136, df = 231, p = 0.892), indicating a decoupled gray matter–only effect. tIV18 was characterized by increased reduced GMC in adolescents with aBPD within medial fronto-temporo-parietal regions and the cerebellum, as well as decreased lateral fronto-parietal GMC. tIV05 showed significant group differences in both imaging modalities, consistent with a multimodal gray–white matter coupled pattern. Gray matter loadings differed significantly between groups (t = 3.488, df = 231, p < 0.001). Importantly, a significant group difference was also observed for the corresponding white matter loadings (t = − 3.165, df = 231, p = 0.002). Spatially, tIV05 was characterized by increased gray matter concentration in adolescents with aBPD within temporal regions, anterior cingulate cortex, and temporal poles, including the amygdala and hippocampus. Moreover, tIV05 showed also a reduction in frontal regions and in the precuneus. For what concerns the WM counterpart, WM portions adjacent to temporo-polar, parietal and cerebellar regions showed a decreased, whereas regions adjacent to the basal ganglia, precuneus and frontal regions displayed a reduction coherently. None of the significant tIV components differed as a function of gender, nor were their loading coefficients significantly correlated with age or years of education (all p > 0.1), indicating that the observed group effects were not driven by demographic variables. See Tables 1 A-B, 2 A-B, 3 A-B, and Figs. 1 , 2 , 3 . tIV05 was characterized by increased gray matter concentration in adolescents with aBPD within temporal regions, anterior cingulate cortex, and temporal poles and by a reduction in frontal regions and in the precuneus. tIV displayed also an increase in GMC in the amygdala, here 3D reconstructed (top). For what concerns the WM counterpart (bottom), WM portions adjacent to temporo-polar, parietal and cerebellar regions showed a decreased, whereas regions adjacent to the basal ganglia, precuneus and frontal regions displayed a reduction coherently. Table 1 A – tIV18 tIV18 – increased GMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Declive * 10.2/10.4 8.8 (-43, -68, -18)/8.8 (24, -82, -15) Lingual Gyrus 17, 18 1.8/2.3 6.6 (-15, -85, -13)/8.5 (19, -84, -13) Fusiform Gyrus 18, 19, 20, 37 2.6/2.4 6.6 (-25, -87, -16)/8.3 (21, -79, -14) Uvula * 1.3/2.4 5.8 (-30, -64, -23)/6.8 (33, -80, -23) Culmen * 7.0/5.8 6.3 (-46, -51, -28)/5.7 (43, -52, -24) Cerebellar Tonsil * 2.8/2.9 6.2 (-49, -51, -32)/5.3 (49, -54, -34) Pyramis * 1.0/2.4 5.4 (-48, -64, -32)/5.0 (34, -74, -33) Inferior Semi-Lunar Lobule * 2.5/3.1 5.0 (-45, -66, -35)/4.9 (22, -76, -38) Inferior Occipital Gyrus 17, 18 0.2/0.3 4.1 (-24, -88, -12)/5.0 (37, -81, -14) Sub-Gyral * 0.1/0.0 4.8 (-3, -81, -9)/-999.0 (0, 0, 0) Middle Occipital Gyrus * 0.0/0.1 -999.0 (0, 0, 0)/4.7 (24, -86, -10) Precuneus * 0.1/0.0 3.7 (0, -71, 46)/-999.0 (0, 0, 0) Declive of Vermis * 0.0/0.1 -999.0 (0, 0, 0)/3.6 (3, -72, -12) Superior Frontal Gyrus 6 0.0/0.1 -999.0 (0, 0, 0)/3.5 (4, 7, 55) Medial Frontal Gyrus * 0.0/0.1 -999.0 (0, 0, 0)/3.5 (1, 41, 39) Table 1 B – tIV18 tIV18 – Decreased GMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Sub-Gyral * 0.1/1.3 3.8 (-55, -48, -5)/6.5 (33, -72, 26) Precentral Gyrus 9 0.0/0.4 -999.0 (0, 0, 0)/5.9 (42, 6, 37) Middle Occipital Gyrus 19 0.0/0.3 -999.0 (0, 0, 0)/5.8 (31, -78, 20) Superior Occipital Gyrus 19 0.0/0.2 -999.0 (0, 0, 0)/5.4 (33, -76, 27) Precuneus * 0.0/0.3 -999.0 (0, 0, 0)/5.2 (28, -75, 20) Middle Frontal Gyrus 8, 9, 10 0.1/0.7 3.9 (-21, 62, 8)/5.2 (39, 8, 40) Inferior Frontal Gyrus 6, 9, 46 0.1/0.7 3.7 (-52, 31, 3)/4.9 (49, 6, 33) Middle Temporal Gyrus 21, 37 1.1/0.4 4.7 (-58, -41, -8)/4.8 (31, -72, 20) Cuneus 30 0.0/0.2 -999.0 (0, 0, 0)/4.7 (21, -68, 10) Posterior Cingulate 30 0.0/0.1 -999.0 (0, 0, 0)/4.4 (24, -65, 9) Inferior Parietal Lobule 40 0.6/0.3 4.2 (-50, -37, 24)/4.4 (52, -29, 24) Insula 13, 40 0.1/0.4 3.6 (-48, -37, 21)/4.2 (50, -25, 18) Postcentral Gyrus 40 0.0/0.2 -999.0 (0, 0, 0)/4.1 (53, -27, 21) Inferior Temporal Gyrus 20, 37 0.2/0.0 4.1 (-46, -2, -34)/-999.0 (0, 0, 0) Lateral Ventricle * 0.0/0.1 -999.0 (0, 0, 0)/3.9 (27, -62, 9) Supramarginal Gyrus * 0.0/0.1 -999.0 (0, 0, 0)/3.8 (50, -47, 27) Superior Frontal Gyrus 10 0.1/0.1 3.5 (-28, 54, -3)/3.7 (24, 58, -8) Angular Gyrus 39 0.0/0.1 -999.0 (0, 0, 0)/3.6 (36, -76, 30) Superior Temporal Gyrus 41 0.0/0.1 -999.0 (0, 0, 0)/3.5 (46, -27, 15) Table 2 A - tIV05 tIV05 – Increased GMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Amygdala 20, 28, 36, 38 2.8/3.1 8.0 (-24, 4, -34)/8.3 (24, 0, -34) Superior Temporal Gyrus 22, 38, 42 3.7/3.8 7.6 (-25, 9, -33)/7.6 (24, 9, -32) Cerebellar Tonsil * 2.6/1.0 7.4 (-12, -55, -41)/5.5 (15, -56, -41) Sub-Gyral * 0.4/0.4 6.8 (-28, 3, -35)/7.2 (27, 3, -34) Middle Temporal Gyrus 20, 21, 38 2.7/0.9 7.0 (-42, 12, -32)/5.9 (33, 8, -37) Inferior Semi-Lunar Lobule * 1.0/0.3 6.8 (-3, -57, -41)/5.0 (4, -60, -41) Inferior Temporal Gyrus 20 1.1/0.2 6.5 (-31, -3, -39)/4.2 (31, -8, -36) Inferior Parietal Lobule 40 0.0/3.2 -999.0 (0, 0, 0)/6.1 (61, -40, 24) Thalamus * 2.4/2.9 5.6 (-9, -13, 8)/5.8 (10, -11, 9) Supramarginal Gyrus 40 0.0/1.1 -999.0 (0, 0, 0)/5.6 (59, -42, 33) Cuneus 30 0.3/0.0 5.2 (-10, -59, 7)/-999.0 (0, 0, 0) Parahippocampal Gyrus 19, 30, 35, 36, 37 1.4/1.3 4.9 (-31, -17, -24)/5.2 (33, -17, -26) Culmen * 1.8/2.0 4.5 (-6, -61, -3)/5.0 (19, -51, -10) Uvula of Vermis * 0.1/0.1 5.0 (-1, -61, -31)/4.1 (3, -61, -34) Posterior Cingulate 30 1.4/0.1 4.9 (-10, -59, 11)/3.7 (19, -51, 7) Lingual Gyrus 18, 19 1.0/0.1 4.8 (-10, -57, 4)/3.8 (1, -86, -6) Culmen of Vermis * 0.1/0.1 4.7 (-1, -63, -4)/4.4 (3, -63, -6) Declive * 0.8/0.3 4.5 (-19, -56, -11)/4.7 (19, -53, -14) Extra-Nuclear * 0.3/0.1 4.5 (-21, -54, 7)/4.1 (9, -4, 4) Postcentral Gyrus 2, 40 0.1/0.1 4.5 (-53, -20, 33)/3.7 (55, -34, 49) Fusiform Gyrus 19, 20 0.6/0.1 4.3 (-21, -56, -7)/3.7 (27, -36, -16) Uvula * 0.2/0.1 4.2 (-4, -64, -31)/4.0 (1, -61, -28) Table 2 B - tIV05 tIV05 – Decreased GMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Inferior Semi-Lunar Lobule * 2.7/1.5 5.2 (-30, -75, -39)/5.8 (24, -72, -42) Middle Occipital Gyrus 18, 19 0.0/1.1 -999.0 (0, 0, 0)/5.5 (30, -85, 10) Pyramis * 1.2/0.1 4.7 (-19, -82, -34)/3.6 (37, -69, -33) Inferior Frontal Gyrus 10, 45, 46, 47 0.5/1.0 4.2 (-49, 27, -5)/4.4 (50, 24, -6) Uvula * 0.1/0.0 4.3 (-10, -80, -34)/-999.0 (0, 0, 0) Cingulate Gyrus 31 1.0/0.5 4.3 (0, -42, 30)/4.1 (3, -39, 31) Middle Frontal Gyrus * 0.4/0.1 4.3 (-40, 35, 22)/3.7 (33, 38, 23) Precuneus 31 0.6/0.0 4.3 (-9, -46, 33)/-999.0 (0, 0, 0) Inferior Occipital Gyrus * 0.0/0.1 -999.0 (0, 0, 0)/3.9 (34, -75, -6) Cerebellar Tonsil * 0.2/0.2 3.9 (-40, -62, -41)/3.9 (42, -60, -42) Postcentral Gyrus * 0.1/0.0 3.8 (-55, -19, 49)/-999.0 (0, 0, 0) Table 3 A – tIV05 tIV05 – Increased WMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Lentiform Nucleus * 0.9/3.3 4.8 (-12, 0, 4)/9.6 (12, 0, 1) Extra-Nuclear * 2.5/7.1 6.2 (-3, -36, 14)/9.0 (13, -1, 6) Thalamus * 0.1/0.9 3.9 (-9, -6, 6)/7.6 (10, -4, 6) Inferior Frontal Gyrus 9, 13, 44, 45, 46, 47 3.6/4.3 7.3 (-42, 28, 8)/7.6 (49, 27, 14) Sub-Gyral * 3.3/4.3 5.6 (-42, 15, 19)/6.8 (46, 24, 14) Cuneus 18 0.0/1.2 -999.0 (0, 0, 0)/6.3 (18, -94, 12) Middle Frontal Gyrus 8, 9, 11, 46, 47 1.5/0.8 6.2 (-30, 31, 32)/5.4 (46, 30, 14) Middle Occipital Gyrus 18 0.1/1.9 4.2 (-24, -88, 18)/5.7 (33, -78, 14) Superior Frontal Gyrus 9 0.2/0.1 5.4 (-31, 35, 31)/3.6 (16, 14, 50) Caudate * 0.0/0.2 -999.0 (0, 0, 0)/4.7 (9, 6, 4) Postcentral Gyrus 3 0.2/0.0 4.6 (-52, -20, 34)/-999.0 (0, 0, 0) Insula 13 0.1/0.2 3.9 (-39, 18, 12)/4.5 (40, 18, 12) Middle Temporal Gyrus 39 0.3/0.1 4.4 (-43, -60, 18)/3.5 (58, -47, -3) Superior Parietal Lobule 7 0.0/0.1 -999.0 (0, 0, 0)/4.2 (34, -62, 43) Posterior Cingulate 23 0.3/0.1 4.0 (-3, -32, 22)/3.6 (4, -28, 24) Declive * 0.2/0.0 4.0 (-13, -62, -15)/-999.0 (0, 0, 0) Lateral Ventricle * 0.1/0.2 3.8 (-12, -36, 17)/3.9 (15, -39, 14) Table 3 B – tIV05 tIV05 – Decreased WMC Area Brodmann Area volume (cc) random effects: Max Value (x, y, z) Middle Frontal Gyrus 9 0.1/1.5 3.9 (-42, 32, 22)/8.0 (42, 28, 26) Sub-Gyral * 1.3/0.8 4.8 (-39, -9, -26)/6.7 (39, 27, 24) Middle Temporal Gyrus 21 1.5/0.4 6.1 (-37, -2, -33)/4.1 (40, 1, -30) Angular Gyrus * 0.4/0.0 5.8 (-48, -58, 35)/-999.0 (0, 0, 0) Inferior Parietal Lobule 40 0.8/1.4 4.7 (-48, -58, 39)/5.7 (43, -52, 45) Inferior Temporal Gyrus 20 0.6/0.0 5.7 (-34, -2, -35)/-999.0 (0, 0, 0) Declive * 1.9/1.7 5.3 (-19, -78, -20)/5.6 (13, -78, -19) Fusiform Gyrus 20 0.3/0.0 5.2 (-42, -3, -26)/-999.0 (0, 0, 0) Supramarginal Gyrus 40 0.4/0.9 5.1 (-46, -55, 32)/4.8 (42, -48, 34) Precuneus 7, 19, 31 0.6/0.5 4.7 (-19, -80, 34)/5.0 (24, -74, 35) Pyramis * 0.7/0.8 4.7 (-10, -80, -26)/4.8 (10, -78, -25) Superior Temporal Gyrus 21, 38, 41, 42 0.4/1.2 4.3 (-56, -13, 1)/4.8 (59, -9, -1) Uvula * 1.2/0.7 4.7 (-34, -71, -23)/4.6 (13, -80, -23) Cuneus 7, 18, 19 0.8/0.2 4.7 (-9, -75, 24)/4.3 (22, -74, 31) Tuber * 1.2/0.0 4.7 (-39, -68, -23)/-999.0 (0, 0, 0) Uncus 20 0.1/0.2 4.5 (-36, -9, -30)/3.8 (33, -5, -34) * * 0.0/0.0 -999.0 (0, 0, 0)/-999.0 (0, 0, 0) Precentral Gyrus 9 0.0/0.3 -999.0 (0, 0, 0)/4.4 (39, 13, 34) Cerebellar Tonsil * 0.1/0.1 3.5 (-40, -60, -41)/3.9 (36, -59, -42) Extra-Nuclear * 0.2/0.0 3.8 (-31, -10, 6)/-999.0 (0, 0, 0) Inferior Semi-Lunar Lobule * 0.2/0.1 3.7 (-40, -63, -38)/3.6 (25, -66, -42) Lingual Gyrus * 0.1/0.0 3.7 (-10, -80, -6)/-999.0 (0, 0, 0) Insula 13 0.0/0.1 -999.0 (0, 0, 0)/3.7 (42, -37, 18) Culmen * 0.1/0.0 3.5 (-40, -54, -29)/-999.0 (0, 0, 0) Brain – symptoms correlations Correlational analyses were conducted between the loading coefficients of the significant tIVA components (tIV05 and tIV18) and the clinical questionnaires, controlling for gender, age, and years of education. Component tIV18 was significantly associated with the Nonacceptance of Emotional Responses subscale (DERS-C; Spearman’s ρ = 0.207, p = 0.020). tIV18 did not show significant associations with the SCARED total score (Spearman’s ρ = −0.095, p = 0.292) or with any of the SCARED subscales (all p > 0.08). Component tIV05 showed a significant negative association with the Emotional Awareness subscale of the DERS (DERS-A; Spearman’s ρ = −0.241, p = 0.007), indicating that higher expression of this network was related to greater difficulties in emotional awareness. tIV18 showed a significant association with Trail Making Test part A (TMT-A) (Spearman’s ρ = −0.233, p < 0.001), indicating that higher expression of this component was associated with poorer processing speed and attentional performance. In addition, tIV18 was significantly negatively correlated with duration of illness (Spearman’s ρ = −0.628, p = 0.019). No significant correlations were observed between tIV18 and remaining questionnaires (all p > 0.05). Component tIV05 showed a significant negative association with overall anxiety severity, as measured by the SCARED total score (Spearman’s ρ = −0.198, p = 0.026). In addition, tIV05 was significantly negatively correlated with the School Phobia subscale of the SCARED (SCARED-D; Spearman’s ρ = −0.198, p = 0.026). tIV05 showed a significant negative association with the SSI-A subscale (Spearman’s ρ = −0.223, p = 0.012), indicating that higher expression of this component was associated with greater severity of self-injurious attitudes. In addition, tIV05 was strongly negatively correlated with SSI-5 (Spearman’s ρ = −0.392, p < 0.001). Component tIV05 showed a significant negative association with the A-DES-II-A subscale (Spearman’s ρ = −0.323, p = 0.035), indicating that higher expression of this component was related to greater dissociative symptoms in this domain. No significant correlations were observed between tIV05 and the A-DES-II total score (Spearman’s ρ = −0.294, p = 0.056). tIV05 showed a significant positive association with global functioning, as assessed by the Global Assessment of Functioning (GAF) scale (Spearman’s ρ = 0.257, p = 0.004), indicating that higher expression of this component was related to better overall psychological, social, and occupational functioning. tIV05 showed a significant negative association with PANSS item 5 (Spearman’s ρ = −0.222, p = 0.037), indicating that higher expression of this component was related to lower severity of this psychopathological dimension. In addition, tIV05 was significantly negatively correlated with Stroop condition 1 performance (Spearman’s ρ = −0.196, p = 0.028). No significant correlations were observed between tIV05 and remaining questionnaires (all p > 0.05). Decision tree To assess the predictive value of the identified structural networks at the individual level, a supervised decision tree classification was implemented to discriminate adolescents with BPD from healthy controls. The model was trained on 149 participants, validated on 38, and tested on an independent hold-out sample of 46 subjects, yielding a test accuracy of 71.7%, with comparable performance across classes. The model achieved an area under the ROC curve (AUC) of 0.687 (see Fig. 4 ), indicating moderate but reliable discriminative ability. Classification performance was higher for the BPD class, with a precision of 0.80, recall of 0.774, and F1 score of 0.787, compared to controls (precision = 0.563; recall = 0.600; F1 = 0.581). The Matthews correlation coefficient (MCC = 0.368) further supported above-chance classification while accounting for class imbalance. Feature importance analysis highlighted the gray matter component tIV18-GM as the most influential predictor, showing the highest relative importance (20.8%) and the largest mean dropout loss, followed by white matter components corresponding to frontal and fronto-limbic pathways (tIV05-WM and tIV05-WM), thus confirming t-tests results but also extending to other components. Notably, the gray matter affective component tIV05 also contributed to classification, albeit with lower relative importance, suggesting a complementary role in refining group separation. The structure of the decision tree (see Fig. 4 ), further underscored the central role of the medial fronto-parietal–cerebellar network (tIV18), which constituted the root node of the model. Initial splits based on tIV18 loadings effectively separated a large proportion of BPD and control participants, with subsequent refinement driven by white matter components, particularly those involving frontal and temporo-limbic connections. This hierarchical organization indicates that gray matter alterations provide the primary axis of discrimination, while white matter features modulate classification at subsequent levels, consistent with a network-based model in which macrostructural alterations constrain, but do not fully determine, diagnostic status. Overall, these results demonstrate that a limited set of data-driven structural brain networks—especially the frontal–cerebellar gray matter component and its associated white matter pathways—can support individual-level classification of adolescent BPD with clinically meaningful accuracy. Importantly, the convergence between unsupervised (tIVA) and supervised (decision tree) analyses reinforces the robustness and translational relevance of the identified networks as potential neurobiological markers of borderline pathology in adolescence. Discussion The present study extends our previous work on structural brain alterations in adolescents with borderline personality disorder (aBPD) by adopting a more advanced, multimodal data-fusion approach that jointly models gray and white matter covariance patterns. Whereas our earlier study relied on source-based morphometry applied exclusively to gray matter, the current investigation leverages transposed Independent Vector Analysis (tIVA) to disentangle both modality-specific and coupled GM–WM networks. This allowed us to refine and expand previous findings, providing a more comprehensive neurobiological account of emotional and behavioral dysregulation in adolescent BPD. Consistent with our prior hypotheses and results, we identified two large-scale neuroanatomical systems that reliably differentiated aBPD from healthy controls: a medial fronto-parietal–cerebellar gray matter network (tIV18) and a multimodal affective GM–WM network (tIV05). Importantly, these networks showed dissociable patterns of clinical association, suggesting that distinct structural systems underlie different dimensions of borderline psychopathology during adolescence. In the next section we discuss the results in detail. Medial fronto-parietal–cerebellar alterations and deficits in emotion regulation and cognitive control (tIV18) In our previous SBM study, one of the most robust findings was the involvement of the cerebellum and posterior default mode network (DMN) hubs, which showed increased gray matter concentration in adolescents with BPD and were associated with difficulties in emotion regulation, self-harm, anxiety symptoms, and reduced global functioning. These results converged with a growing body of evidence indicating that the cerebellum plays a critical role in emotional regulation, anxiety, depression, and self-injurious behaviors, beyond its classical motor functions (Salvador et al., 2011 ; Koziol et al., 2014 ; Hallett et al., 2016 ). In adult BPD, posterior cerebellar abnormalities have similarly been linked to impulsivity and emotional dysregulation, and cerebellar–limbic interactions have been proposed as mechanisms underlying maladaptive coping strategies such as non-suicidal self-injury (Schulze et al., 2016 ; Koziol et al., 2014 ; Dadomo et al., 2018 ). The present findings partially converge with, but also refine, this interpretation. The tIV18 component encompasses medial prefrontal, posterior midline, and cerebellar regions that together map onto a medial fronto–cerebellar loop supporting self-referential monitoring, emotional appraisal, and cognitive efficiency. Within contemporary models of cerebellar function, such loops are thought to provide predictive modulation and error signaling to prefrontal systems, thereby constraining emotional responses and internal evaluative processes. Crucially, tIV18 exhibited a gray matter–only pattern, with no corresponding white matter alterations, suggesting that during adolescence structural deviations within medial fronto-parietal–cerebellar systems may primarily reflect regional morphological alterations rather than disrupted long-range connectivity. Functionally, tIV18 was selectively associated with nonacceptance of emotional responses, a core dimension of emotion dysregulation in BPD. This association is consistent with the role of medial prefrontal and posterior midline regions, key hubs of the default mode network (DMN), in self-referential and evaluative processing, including internal dialogue and affective appraisal (Menon, 2011; Grecucci et al., 2022 ). Alterations within these systems have been linked to distorted self-related processing and maladaptive internal evaluation, hallmark features of borderline pathology, and DMN hyperactivity at rest has been repeatedly associated with difficulties in affective and interpersonal regulation (Koenigsberg et al., 1999 ; Ruocco et al., 2013 ). From a structural perspective, integrity of posterior DMN regions has also been related to the capacity to engage adaptive emotion regulation strategies such as cognitive reappraisal (Ghoumroudi et al., 2023). Compared to our previous study, the clinical association profile of tIV18 appears more circumscribed: rather than correlating broadly with anxiety and self-harm, this network was specifically linked to emotional nonacceptance, attentional efficiency (as indexed by TMT-A), and illness duration. This pattern suggests that medial fronto-parietal–cerebellar alterations may index a more fundamental vulnerability related to emotional monitoring and cognitive efficiency, potentially shaping illness course rather than momentary symptom severity. Taken together, we put forward the hypothesis that adolescent BPD reflects a dysfunction of fronto–cerebellar loops that normally orchestrate adaptive monitoring and modulation of internal emotional states, rather than being reducible to limbic hyperreactivity alone. We propose that early structural deviations within these circuits disrupt the predictive calibration of emotional responses and the tolerance of aversive affect, thereby fostering persistent emotional nonacceptance and a rigid, maladaptive style of self-regulation that is central to borderline pathology. Affective GM–WM network alterations and emotional–behavioral dysregulation (tIV05) The second component, tIV05, represents a key methodological and clinical advance by demonstrating that anxiety, self-injurious behaviors, and functional impairment in adolescent BPD are more coherently captured by a coupled gray–white matter affective network rather than by gray matter alterations alone. This network encompasses temporal regions, anterior cingulate cortex, temporal poles, and limbic structures such as the amygdala and hippocampus, together with convergent white matter alterations in frontal, parietal, and temporo-limbic pathways. Importantly, this configuration closely maps onto affective–limbic circuits central to emotional reactivity and affect regulation in BPD. Functional neuroimaging studies have consistently reported exaggerated limbic responses to negative emotional stimuli in BPD, including heightened activation of the amygdala, hippocampus, parahippocampal regions, superior frontal gyrus, and cerebellum (LeBoeuf et al., 2016 ), while structural and functional abnormalities within these circuits have been linked to anxiety, heightened emotional sensitivity, and self-injurious behaviors across adolescent and adult samples (Schulze et al., 2016 ; Quattrini et al., 2022 ; Langerbeck et al., 2023 ). The present findings extend this literature by showing that such affective disturbances are supported by network-level GM–WM dyscoordination rather than by isolated regional alterations. Clinically, tIV05 was associated with greater anxiety severity, school-related anxiety, self-injurious attitudes and behaviors, and dissociative symptoms, while also showing a positive association with global functioning (GAF). This pattern suggests that tIV05 indexes a core affective liability in adolescent BPD that directly relates to clinically relevant dimensions of emotional and behavioral dysregulation, including self-harm risk. At the same time, its association with global functioning indicates that heightened affective sensitivity may coexist with relatively preserved functioning depending on contextual and regulatory factors, consistent with the marked heterogeneity observed in adolescent BPD. Crucially, the involvement of white matter within tIV05 highlights disrupted communication between limbic emotional generators and frontal regulatory systems as a central neurobiological substrate of affective dysregulation in aBPD. This finding strengthens fronto-limbic disconnection models of BPD by showing that anxiety, dissociation, and self-harm are more tightly linked to impaired structural integration than to gray matter abnormalities alone, in line with adult evidence of altered neurodevelopmental and myelination trajectories in affective and executive networks (Quattrini et al., 2022 ). Overall, tIV05 emerges as a clinically meaningful affective network closely tied to the dimensions that most strongly drive clinical severity, risk, and treatment need in adolescent BPD. A model for aBPD and its implications In our previous study, reduced gray matter concentration in lateral prefrontal regions,corresponding to the central executive network (CEN), was interpreted as reflecting deficits in top-down control and impulse regulation in adolescent BPD. This interpretation was supported by converging structural and functional evidence of dorsolateral prefrontal cortex abnormalities in both adolescents and adults with BPD (Hazlett et al., 2005 ; Brunner et al., 2010 ; Bozzatello et al., 2021 ), as well as functional hypometabolism in prefrontal regions implicated in cognitive control (Goethals et al., 2005 ; Soloff et al., 2003 ). The present findings refine this view by showing that prefrontal alterations do not emerge as an isolated deficit, but rather as components of dissociable large-scale networks involving medial fronto-parietal–cerebellar regions (tIV18) and a coupled affective GM–WM system (tIV05). From a translational perspective, this network-level organization suggests that emotion dysregulation in adolescent BPD reflects an imbalance between affective drive and regulatory control, rather than a uniform impairment of executive functioning. The affective GM–WM network (tIV05), which is strongly associated with anxiety, self-injurious behaviors, dissociative symptoms, and functional outcome, may represent a primary target for interventions aimed at attenuating maladaptive emotional reactivity, for instance through neuromodulatory approaches targeting prefrontal nodes connected to limbic circuits. Conversely, the medial fronto-parietal–cerebellar network (tIV18), associated with emotional nonacceptance and reduced cognitive efficiency, highlights a complementary target for interventions aimed at strengthening executive and attentional control processes. Importantly, conceptualizing these alterations at the network level suggests that effective early interventions in adolescent BPD may need to be functionally differentiated and network-specific, targeting affective hyper-responsivity and regulatory control through distinct but complementary mechanisms. This perspective is particularly relevant in adolescence, a developmental period marked by heightened neuroplasticity and ongoing maturation of fronto-parietal control systems. Limitations and conclusions This study provides robust evidence that adolescent borderline personality disorder is characterized by distinct large-scale structural brain networks, rather than isolated regional abnormalities. By adopting a multimodal, data-driven framework, our findings point to fronto–cerebellar loop dysfunction as a previously underexplored neurodevelopmental mechanism contributing to emotion dysregulation and self-injurious vulnerability in adolescent borderline personality disorder. Several considerations should be taken into account when interpreting these findings. Participants were recruited from a tertiary psychiatric center and predominantly represented adolescents with moderate-to-severe BPD, characterized by high levels of emotion dysregulation, frequent non-suicidal self-injury, comorbid anxiety, and functional impairment. While this may limit generalizability to milder or subthreshold presentations, it also constitutes a strength of the study, as it ensures high diagnostic reliability and clinical relevance. Exclusion of major psychiatric comorbidities known to exert strong, independent effects on brain structure enhanced internal validity and allowed for a more precise characterization of BPD-related neural alterations. Although comorbid anxiety remained prevalent, this reflects the typical clinical profile of adolescent BPD and underscores the ecological validity of the sample. Finally, while the present study focused on structural MRI, future research should extend this multimodal framework to include functional and longitudinal measures to better characterize developmental trajectories and treatment-related changes. Declarations Conflict of interest The authors declare that they have no conflicts of interest Ethical approval All procedure complied with the ethical standard of the 1964 Helsinki Declaration Institutional Review Board approval was obtained from our Research Ethics Committee (IRB: 2022020227) Written informed consent was obtained from legal guardians of all adolescent participants Funding None. Author Contribution AG analyzed the data, visualized the results, and wrote the manuscriptXY, QX conceived this project, wrote and revised the manuscriptXL, XW, JZ, FW, LS, LW wrote and revised the manuscript Acknowledgment The authors have no acknowledgements to declare Data Availability The data that support the findings of this study are available on request from the corresponding authors. The data are not publicly available due to their containing information that could compromise the privacy of research participants References Aas, I. H. M. (2011). Guidelines for rating Global Assessment of Functioning (GAF). Annals of General Psychiatry, 10 (1), 2. https://doi.org/10.1186/1744-859X-10-2 Adalı, T., & Calhoun, V. D. (2015). Multimodal data fusion: An overview of methods, challenges, and prospects. Proceedings of the IEEE, 103(9), 1449–1477. Alexander-Bloch, A., Giedd, J. N., & Bullmore, E. (2013). Imaging structural co-variance between human brain regions. Nature Reviews Neuroscience, 14 (5), 322–336. https://doi.org/10.1038/nrn3465 Ashburner, J., & Friston, K. J. (2000). Voxel-based morphometry—the methods. NeuroImage, 11 (6), 805–821. Baggio, T., Grecucci, A., Meconi, F., & Messina, I. (2023). Anxious brains: A combined data fusion machine learning approach to predict trait anxiety from morphometric features. Sensors, 23 (2). Bassett, D. S., Bullmore, E., Verchinski, B. A., Mattay, V. S., Weinberger, D. R., & Meyer-Lindenberg, A. (2008). Hierarchical organization of human cortical networks in health and schizophrenia. Journal of Neuroscience, 28 (37), 9239–9248. https://doi.org/10.1523/JNEUROSCI.1929-08.2008 Birmaher, B., Brent, D. A., Chiappetta, L., Bridge, J., Monga, S., & Baugher, M. (1999). Psychometric properties of the Screen for Child Anxiety Related Emotional Disorders (SCARED): A replication study. Journal of the American Academy of Child & Adolescent Psychiatry, 38 (10), 1230–1236. Biswal, B. B., Yetkin, F. Z., Haughton, V. M., & Hyde, J. S. (1995). Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magnetic Resonance in Medicine, 34(4), 537–541. Bohus, M., et al. (2021). Borderline personality disorder. The Lancet, 398 (10310), 1528–1540. Bozzatello, P., Morese, R., Valentini, M. C., Rocca, P., & Bellino, S. (2021). How interpersonal psychotherapy changes the brain: A study of fMRI in borderline personality disorder. Journal of Clinical Psychiatry, 83 (1). https://doi.org/10.4088/JCP.21m13918 Brunner, R., et al. (2010). Reduced prefrontal and orbitofrontal gray matter in female adolescents with borderline personality disorder: Is it disorder-specific? NeuroImage, 49 (1), 114–120. Calhoun, V. D., Adalı, T., Pearlson, G. D., & Pekar, J. J. (2006). A method for making group inferences from functional MRI data using independent component analysis. Human Brain Mapping, 14(3), 140–151. Canessa, N., Crespi, C., Motterlini, M., Baud-Bovy, G., Chierchia, G., Pantaleo, G., et al. (2013). The functional and structural neural basis of individual differences in social behavior. Journal of Neuroscience, 33 (36), 14307–14317. Chanen, A. M., et al. (2008). Orbitofrontal, amygdala, and hippocampal volumes in teenagers with first-presentation borderline personality disorder. Psychiatry Research: Neuroimaging, 163 (2), 116–125. Dadomo, H., Grecucci, A., Giardini, I., Ugolini, E., Carmelita, A., & Panzeri, M. (2016). Schema therapy for emotional dysregulation: Theoretical implications and clinical application. Frontiers in Psychology, 7 , 1987. https://doi.org/10.3389/fpsyg.2016.01987 Dadomo, H., et al. (2018). Neurobiological underpinnings of borderline personality disorder. Psychiatric Research. Dadomo, H., Panzeri, M., Caponcello, D., Carmelita, A., & Grecucci, A. (2018). Schema therapy for emotional dysregulation in personality disorders: A review. Current Opinion in Psychiatry, 31 (1), 43–49. Dadomo, H., Salvato, G., Lapomarda, G., Ciftci, Z., Messina, I., & Grecucci, A. (2022). Structural features predict sexual trauma and interpersonal problems in borderline personality disorder but not in controls: A multi-voxel pattern analysis. Frontiers in Human Neuroscience, 23 . Depping, M. S., Wolf, N. D., Vasic, N., Sambataro, F., Thomann, P. A., & Wolf, R. C. (2016). Common and distinct structural network abnormalities in major depressive disorder and bipolar disorder. Progress in Neuropsychopharmacology and Biological Psychiatry, 65 , 127–133. Ding, J. B., & Hu, K. (2021). Structural MRI brain alterations in borderline personality disorder and bipolar disorder. Cureus, 13 , e16425. https://doi.org/10.7759/cureus.16425 Ellison, W. D., et al. (2018). Community and clinical epidemiology of borderline personality disorder. Psychiatric Clinics of North America, 41 (4), 561–573. Evans, A. C. (2013). Networks of anatomical covariance. NeuroImage, 80 , 489–504. https://doi.org/10.1016/j.neuroimage.2013.05.054 Fettes, P., Schulze, L., & Downar, J. (2017). Cortico-striatal-thalamic loop circuits of the orbitofrontal cortex: Promising therapeutic targets in psychiatric illness. Frontiers in Systems Neuroscience, 11 , 25. https://doi.org/10.3389/fnsys.2017.00025 Fox, M. D., Snyder, A. Z., Vincent, J. L., Corbetta, M., Van Essen, D. C., & Raichle, M. E. (2005). The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the USA, 102(27), 9673–9678. doi:10.1073/pnas.0504136102 Frederickson, J. J., Messina, I., & Grecucci, A. (2018). Dysregulated anxiety and dysregulating defenses: Toward an emotion regulation informed dynamic psychotherapy. Frontiers in Psychology, 9 , 2054. https://doi.org/10.3389/fpsyg.2018.02054 Ghomroudi, P. A., Scaltritti, M., & Grecucci, A. (2023). Decoding reappraisal and suppression from neural circuits: A combined supervised and unsupervised machine learning approach. Cognitive, Affective, & Behavioral Neuroscience, 23 , 1095–1112. https://doi.org/[Insert DOI] Giuliani, N. R., Drabant, E. M., & Gross, J. J. (2011). Anterior cingulate cortex volume and emotion regulation: Is bigger better? Biological Psychology, 86 (3), 379–382. https://doi.org/10.1016/j.biopsycho.2010.11.010 Goethals, I., Audenaert, K., Jacobs, F., Van den Eynde, F., Bernagie, K., Kolindou, A., ... & Van Heeringen, C. (2005). Brain perfusion SPECT in impulsivity-related personality disorders. Behavioural Brain Research, 157 (1), 187-192. https://doi.org/10.1016/j.bbr.2004.06.022 Goodman, M., Hazlett, E. A., Avedon, J. B., Siever, D. R., Chu, K. W., & New, A. S. (2011). Anterior cingulate volume reduction in adolescents with borderline personality disorder and co-morbid major depression. Journal of Psychiatric Research, 45 (6), 803-807. https://doi.org/10.1016/j.jpsychires.2010.11.011 Grecucci, A., Dadomo, H., Salvato, G., Lapomarda, G., Sorella, S., & Messina, I. (2023). Abnormal brain circuits characterize borderline personality and mediate the relationship between specific childhood traumas and symptoms: An mCCA+ jICA and random forest approach. Sensors, 23 (5). Grecucci, A., Lapomarda, G., Messina, I., Monachesi, B., Sorella, S., & Siugzdaite, R. (2022). Structural features related to affective instability correctly classify the diagnosis of borderline personality disorder: A supervised machine learning approach. Frontiers in Psychiatry . https://doi.org/10.3389/fpsyt.2022.804440 Groves, A. R., Beckmann, C. F., Smith, S. M., & Woolrich, M. W. (2011). Linked independent component analysis for multimodal data fusion. NeuroImage, 54(3), 2198–2217. doi:10.1016/j.neuroimage.2010.09.073 Hallett, M., et al. (2016). Cerebellar contributions to emotional processing. Cerebellum . https://doi.org/[Insert DOI] Hazlett, E. A., New, A. S., Newmark, R., Haznedar, M. M., Lo, J. N., Speiser, L. J., ... & Buchsbaum, M. S. (2005). Reduced anterior and posterior cingulate gray matter in borderline personality disorder. Biological Psychiatry, 58 (8), 614-623. https://doi.org/10.1016/j.biopsych.2005.04.029 Hebart, M. N., & Baker, C. I. (2018). Deconstructing multivariate decoding for the study of brain function. NeuroImage, 180, 4–18. Hyvärinen, A., & Oja, E. (2000). Independent component analysis: Algorithms and applications. Neural Networks, 13 (4–5), 411–430. Irle, E., et al. (2007). Size abnormalities of the superior parietal cortices are related to dissociation in borderline personality disorder. Psychiatry Research, 156 (2), 139–149. Jin, X., et al. (2016). A voxel-based morphometric MRI study in young adults with borderline personality disorder. PLOS ONE, 11 (1), e0147938. Kherif, F., & Latypova, A. (2020). Principal component analysis. In Mechelli & Vieira (Eds.), Machine Learning: Methods and Applications to Brain Disorders (Academic Press). Koenigsberg, H. W., Anwunah, I., New, A. S., Mitropoulou, V., Schopick, F., & Siever, L. J. (1999). Relationship between depression and borderline personality disorder. Depression and Anxiety, 10 , 158–167. https://doi.org/10.1002/(SICI)1520-6394(1999)10:43.0.CO;2-B Koziol, L. F., et al. (2014). The cerebellum and emotion: A review of recent findings. Neuropsychology Review . Langerbeck, M., Baggio, T., Messina, I., Bhat, S., & Grecucci, A. (2023). Borderline shades: Morphometric features predict borderline personality traits but not histrionic traits. NeuroImage: Clinical, 40 , 103530. https://doi.org/10.1016/j.nicl.2023.103530 Lapomarda, G., Grecucci, A., Messina, I., Pappaianni, E., & Dadomo, H. (2021). Common and different gray and white matter alterations in bipolar and borderline personality disorder: A source-based morphometry study. Brain Research, 1762 , 147401. https://doi.org/10.1016/j.brainres.2021.147401 Lapomarda, G., Pappaianni, E., Siugzdaite, R., Sanfey, A. G., Rumiati, R. I., & Grecucci, A. (2021). Out of control: An altered parieto-occipital-cerebellar network for impulsivity in bipolar disorder. Behavioral Brain Research, 406 , 113228. https://doi.org/10.1016/j.bbr.2021.113228 LeBoeuf, A., Guilé, J. M., Labelle, R., & Luck, D. (2016). Functional neuroimaging pilot study of borderline personality disorder in adolescents. Sante Mentale au Quebec, 41 (1), 141–162. Lee, J.-H., Lee, T.-W., Jolesz, F. A., & Yoo, S.-S. (2008). Independent vector analysis (IVA): Multivariate approach for fMRI group study. NeuroImage, 40(1), 86–109. doi:10.1016/j.neuroimage.2007.11.019 Levine, A. Z., et al. (2020). Nonsuicidal self-injury and suicide: Differences between those with and without borderline personality disorder. Journal of Personality Disorders, 34 (1), 131–144. Li, J., et al. (2018). Psychometric properties of the Chinese version of the Difficulties in Emotion Regulation Scale (DERS): Factor structure, reliability, and validity. Psychological Assessment, 30 (5), e1–e9. Luo, X., Mao, Q., Shi, J., Wang, X., & Li, C. R. (2019). Putamen gray matter volumes in neuropsychiatric and neurodegenerative disorders. World Journal of Psychiatry and Mental Health Research, 3 , 1020. Martin, J., et al. (2013). Psychometric properties of the functions and addictive features scales of the Ottawa Self-Injury Inventory: A preliminary investigation using a university sample. Psychological Assessment, 25 (3), 1013–1018. McKeown, M. J., & Sejnowski, T. J. (1998). Independent component analysis of fMRI data: Examining the assumptions. Human Brain Mapping, 6 (5–6), 368–372. Mirkovic, B., et al. (2021). Borderline personality disorder and adolescent suicide attempt: The mediating role of emotional dysregulation. BMC Psychiatry, 21 (1), 393. Pappaianni, E., Siugzdaite, R., Vettori, S., Venuti, P., Job, R., & Grecucci, A. (2018). Three shades of grey: Detecting brain abnormalities in children with autism by using source-, voxel- and surface-based morphometry. European Journal of Neuroscience, 47 (6), 690–700. Paris, J., Gunderson, J., & Weinberg, I. (2007). The interface between borderline personality disorder and bipolar spectrum disorders. Comprehensive Psychiatry, 48 , 145–154. https://doi.org/10.1016/j.comppsych.2006.10.001 Pearlson, G. D., Calhoun, V. D., & Liu, J. (2015). An introductory review of parallel independent component analysis (p-ICA) and a guide to applying p-ICA to genetic data and imaging phenotypes to identify disease-associated biological pathways and systems in common complex disorders. Frontiers in Genetics, 6 , 276. Quattrini, G., Magni, L. R., Lanfredi, M., Pedrini, L., Carcione, A., Riccardi, I., Corbo, D., Gasparotti, R., Rossi, R., Pievani, M., & Caroli, A. (2022). Aberrant structural connectivity of the triple network system in borderline personality disorder is associated with behavioral dysregulation. Journal of Clinical Medicine, 11 (7). https://doi.org/10.3390/jcm11071757 Richter, J., et al. (2014). Reduced cortical and subcortical volumes in female adolescents with borderline personality disorder. Psychiatry Research, 221 (3), 179–186. Rossi, R., Pievani, M., Lorenzi, M., Boccardi, M., Beneduce, R., Bignotti, S., et al. (2013). Structural brain features of borderline personality and bipolar disorders. Psychiatry Research, 213 , 83–91. https://doi.org/10.1016/j.pscychresns.2012.07.002 Ruocco, A. C., Amirthavasagam, S., Choi-Kain, L. W., & McMain, S. F. (2013). Neural correlates of negative emotionality in borderline personality disorder: An activation-likelihood-estimation meta-analysis. Biological Psychiatry, 73 , 153–160. https://doi.org/10.1016/j.biopsych.2012.07.014 Salvador, R., et al. (2011). Cerebellar contributions to anxiety and depression. Neuroscience . Saviola, F., Pappaianni, E., Monti, A., Grecucci, A., Jovicich, J., & de Pisapia, N. (2020). Trait and state anxiety are mapped differently in the human brain. Scientific Reports, 10 (1). https://doi.org/10.1038/s41598-020-68008-z Schnack, H. G. (2020). Improving individual predictions: Machine learning in psychiatry. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 5(6), 560–571. Schmahmann, J. D. (2010). The role of the cerebellum in emotion and cognition. Cerebellum . Schulze, L., Schmahl, C., & Niedtfeld, I. (2016). Neural correlates of disturbed emotion processing in borderline personality disorder: A multimodal meta-analysis. Biological Psychiatry, 79 , 97–106. https://doi.org/10.1016/j.biopsych.2015.03.027 Schulze, L., et al. (2016). The cerebellum and emotion dysregulation in borderline personality disorder. Frontiers in Psychology . Seeley, W. W., Crawford, R. K., Zhou, J., Miller, B. L., & Greicius, M. D. (2009). Neurodegenerative diseases target large-scale human brain networks. Neuron, 62 (1), 42–52. https://doi.org/10.1016/j.neuron.2009.03.024 Soloff, P., Nutche, J., Goradia, D., & Diwadkar, V. (2008). Structural brain abnormalities in borderline personality disorder: A voxel-based morphometry study. Psychiatry Research, 164 , 223–236. https://doi.org/10.1016/j.pscychresns.2008.02.003 Soloff, P. H., Kelly, T. M., Strotmeyer, S. J., Malone, K. M., & Mann, J. J. (2003). Impulsivity, gender, and response to fenfluramine challenge in borderline personality disorder. Psychiatry Research, 119 (1-2), 11–24. https://doi.org/10.1016/s0165-1781(03)00100-8 Sorella, S., Lapomarda, G., Messina, I., Frederickson, J. J., Siugzdaite, R., Job, R., et al. (2019). Testing the expanded continuum hypothesis of schizophrenia and bipolar disorder. Neural and psychological evidence for shared and distinct mechanisms. NeuroImage, 23 , 101854. https://doi.org/10.1016/j.nicl.2019.101854 Sui, J., Adalı, T., Yu, Q., Chen, J., & Calhoun, V. D. (2012). A review of multivariate methods for multimodal fusion of brain imaging data. Journal of Neuroscience Methods, 204(1), 68–81. doi:10.1016/j.jneumeth.2011.10.031 Vieira, S., Pinaya, W. H. L., & Mechelli, A. (2020). Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders. Neuroscience & Biobehavioral Reviews, 113, 546–560. Wechsler, D. (1999). Wechsler abbreviated scale of intelligence . The Psychological Corporation. Whittle, S., Chanen, A. M., Fornito, A., McGorry, P. D., Pantelis, C., & Yücel, M. (2009). Anterior cingulate volume in adolescents with first-presentation borderline personality disorder. Psychiatry Research, 172 (2), 155–160. https://doi.org/10.1016/j.pscychresns.2008.12.004 Wood, A., Kroll, L., Moore, A., & Harrington, R. (1995). Properties of the Mood and Feelings Questionnaire in adolescent psychiatric outpatients: A research note. Journal of Child Psychology and Psychiatry, 36 (2), 327–334. Xiao, Q., Shen, L., He, H., et al. (2024). Alteration of prefrontal cortex and its associations with emotional and cognitive dysfunctions in adolescent borderline personality disorder. European Child and Adolescent Psychiatry, 33 , 3937–3949. https://doi.org/10.1007/s00787-024-02438-2 Xu, L., Groth, K. M., Pearlson, G., Schretlen, D. J., & Calhoun, V. D. (2009). Source-based morphometry: The use of independent component analysis to identify gray matter differences with application to schizophrenia. Human Brain Mapping, 30 (3), 711–724. Xu, T., Cullen, K. R., Mueller, B., Schreiner, M. W., Lim, K. O., Schulz, S. C., et al. (2016). Network analysis of functional brain connectivity in borderline personality disorder using resting-state fMRI. NeuroImage, 11 , 302–315. https://doi.org/10.1016/j.nicl.2016.02.006 Yi, X., Fu, Y., Ding, J., Jiang, F., Han, Z., Zhang, Y., et al. (2024). Altered gray matter volume and functional connectivity in adolescent borderline personality disorder with non-suicidal self-injury behavior. European Child and Adolescent Psychiatry, 33 (1), 193–202. https://doi.org/10.1007/s00787-023-02161-4 Yi, X., Fu, Y., Zhang, Z., Jiang, F., Xiao, Q., & Chen, B. T. (2023). Altered regional homogeneity and its association with cognitive function in adolescents with borderline personality disorder. Journal of Psychiatry and Neuroscience, 48 (1), E1–E10. https://doi.org/10.1503/jpn.220144 Yi, X., Wang, X., Shen, L., Fu, Y., Zhang, Z., Ding, J., Jiang, F., Zhang, Z., Grecucci, A., & Chen, B. T. (2025). White matter microstructural abnormalities related to emotional dysfunction and childhood trauma characterize adolescents with borderline personality disorder . Journal of Affective Disorders. Advance online publication. https://doi.org/10.1007/s11682-025-01046-1 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8860480","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":591006549,"identity":"40412898-2eca-47ea-8a54-f58474f170f3","order_by":0,"name":"Xiaoping Yi","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaoping","middleName":"","lastName":"Yi","suffix":""},{"id":591006551,"identity":"139450b8-47ec-4299-b623-3cb728a7963a","order_by":1,"name":"Xue Liu","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xue","middleName":"","lastName":"Liu","suffix":""},{"id":591006553,"identity":"07c74273-743f-4401-b740-0e28819891ca","order_by":2,"name":"Xueying Wang","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xueying","middleName":"","lastName":"Wang","suffix":""},{"id":591006555,"identity":"5ef420f2-71be-4c8f-860f-e8b8d68e7303","order_by":3,"name":"Jinfan Zhang","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jinfan","middleName":"","lastName":"Zhang","suffix":""},{"id":591006558,"identity":"0def0b19-6e41-418a-8e8e-80acfa0131ea","order_by":4,"name":"Feifei Wu","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Feifei","middleName":"","lastName":"Wu","suffix":""},{"id":591006560,"identity":"060c10b6-af97-4f66-b309-157d970fea17","order_by":5,"name":"Liying Shen","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Liying","middleName":"","lastName":"Shen","suffix":""},{"id":591006561,"identity":"8be35261-fe8b-49ec-82db-6e84a8a5e6bf","order_by":6,"name":"Lin Wang","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lin","middleName":"","lastName":"Wang","suffix":""},{"id":591006563,"identity":"8060ed8a-2ff0-40ae-86fa-b86315af4e91","order_by":7,"name":"Qian Xiao","email":"","orcid":"","institution":"Chongqing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qian","middleName":"","lastName":"Xiao","suffix":""},{"id":591006564,"identity":"648a0e5a-921a-4fb3-ad1b-263c69dfe008","order_by":8,"name":"Alessandro Grecucci","email":"data:image/png;base64,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","orcid":"","institution":"University of Bari Aldo Moro","correspondingAuthor":true,"prefix":"","firstName":"Alessandro","middleName":"","lastName":"Grecucci","suffix":""}],"badges":[],"createdAt":"2026-02-12 09:53:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8860480/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8860480/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104180594,"identity":"26aecbd1-c597-4f5c-b1b0-24a6061bff5f","added_by":"auto","created_at":"2026-03-08 17:16:26","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1020114,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNeural results. \u003c/strong\u003etIV18 was characterized by increased reduced GMC in adolescents with aBPD within medial fronto-temporo-parietal regions and the cerebellum (top), as well as decreased lateral fronto-parietal GMC (bottom).\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8860480/v1/ad340079482dd676ecbd4f85.jpeg"},{"id":104180596,"identity":"c3174127-b065-454f-a48a-27e0110bc843","added_by":"auto","created_at":"2026-03-08 17:16:26","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1050523,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNeural results of tIV05\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003etIV05 was characterized by increased gray matter concentration in adolescents with aBPD within temporal regions, anterior cingulate cortex, and temporal poles and by a reduction in frontal regions and in the precuneus.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8860480/v1/3656449c2fa765773b41da89.jpeg"},{"id":104180597,"identity":"22149e48-a028-4c26-96a8-88c772ab9464","added_by":"auto","created_at":"2026-03-08 17:16:26","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":671554,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNeural results of tIV05, amygdala and WM.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003etIV displayed also an increase in GMC in the amygdala, here 3D reconstructed (top). For what concerns the WM counterpart (bottom), WM portions adjacent to temporo-polar, parietal and cerebellar regions showed a decreased, whereas regions adjacent to the basal ganglia, precuneus and frontal regions displayed a reduction coherently.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8860480/v1/3d892537edf3ce851feed629.jpeg"},{"id":104180595,"identity":"b830d694-96cd-42c8-a8b1-ca9f50bc4455","added_by":"auto","created_at":"2026-03-08 17:16:26","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":327294,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eResults from decision tree classification\u003c/strong\u003e. Decision tree confirmed the role of the main components to classify new adolescents with BPD (Hold out method). Top: ROC, middle: feature importance, bottom: decision tree structure\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8860480/v1/125302aff2cd0832d964b80c.jpeg"},{"id":104779393,"identity":"2d415b7b-84fb-4ea4-b0cf-13c85b3f716d","added_by":"auto","created_at":"2026-03-17 07:39:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4489536,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8860480/v1/57184651-8177-4749-a213-5c1262776717.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Coupled and decoupled fronto–cerebellar and affective brain networks characterize adolescents with borderline personality disorder","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBorderline personality disorder (BPD) represents a highly prevalent and clinically burdensome condition, affecting approximately 1\u0026ndash;6% of the general population, with substantially higher rates reported in outpatient (up to 10%) and inpatient psychiatric settings (up to 25%) (Ellison et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Increasing attention has been devoted to borderline personality disorder in adolescence (aBPD), which is now recognized as a valid and clinically meaningful condition rather than a transient developmental phase. aBPD is primarily characterized by pervasive difficulties in emotion regulation and a high incidence of non-suicidal self-injury (NSSI) (Bohus et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Mirkovic et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Emotion regulation difficulties in aBPD encompass a broad range of deficits, including impaired emotional awareness and understanding, heightened emotional reactivity, reduced capacity to modulate affective responses, and limited access to adaptive regulatory strategies. Clinical observations indicate that adolescents with BPD frequently exhibit several, if not all, of these impairments. Importantly, converging theoretical and empirical evidence suggests that such emotion regulation failures may play a causal role in the emergence and maintenance of NSSI behaviors (Dadomo et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Despite the clinical relevance of aBPD, its neurobiological underpinnings, particularly the neural correlates of emotion dysregulation and NSSI, remain insufficiently characterized. This gap is largely attributable to the fact that the majority of neuroimaging studies on BPD have been conducted in adult populations. Structural MRI investigations in adults with BPD have consistently reported gray matter alterations within limbic\u0026ndash;frontal circuits, including reduced volumes in the bilateral hippocampus, dorsolateral prefrontal cortex, and medial frontal cortex (Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Rossi et al., \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Prossin et al., 2010; Minzenberg et al., 2008; Xu et al., \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Soloff et al., \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Luo et al., 2010; Ding et al., 2021; Grecucci et al., 2021; \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; 2024). These structural abnormalities are functionally meaningful, as they map onto core clinical features of BPD, such as affective instability, impaired impulse control, and disturbed interpersonal functioning (Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Dadomo et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2016\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Frederickson et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Ruocco et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). In addition to limbic and posterior alterations, functional neuroimaging studies have implicated abnormal activity within the default mode network (DMN), particularly increased resting-state engagement, in affective dysregulation and maladaptive self-referential and interpersonal processes in BPD (Ruocco et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Paris et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Koenigsberg et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e1999\u003c/span\u003e). However, despite growing recognition of adolescence as a critical neurodevelopmental window for the emergence of BPD-related pathology, only a limited number of studies have specifically examined the neural architecture of aBPD.\u003c/p\u003e \u003cp\u003eAdolescence represents a critical neurodevelopmental period characterized by profound structural and functional brain reorganization. Investigating borderline personality disorder during this developmental window is therefore essential to improve early detection and to inform timely and targeted interventions. Neuroimaging studies focusing on adolescents with BPD have provided initial evidence of gray matter abnormalities within limbic and frontal circuits, including alterations in the bilateral orbitofrontal and dorsolateral prefrontal cortices, right amygdala, left anterior cingulate cortex, and bilateral hippocampus (Chanen et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Richter et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Brunner et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Goodman et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Yi et al., \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Nevertheless, the involvement of large-scale networks such as the default mode network (DMN) and the precise contribution of prefrontal control regions in aBPD remain insufficiently established. Beyond cortico-limbic regions, a growing body of evidence has highlighted the cerebellum as a key node in affective and cognitive regulation across a range of psychiatric conditions (Salvador et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Koziol et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Hallett et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Importantly, cerebellar contributions to emotion and cognition are thought to be implemented through recurrent frontal\u0026ndash;cerebellar loops linking medial and lateral prefrontal regions with posterior cerebellar territories, supporting predictive control, error monitoring, and adaptive regulation of internal states (e.g., Schmahmann, 2019; Stoodley \u0026amp; Schmahmann, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2010\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWithin this framework, the cerebellum is not conceived as an isolated affective structure, but as a modulatory node embedded in fronto\u0026ndash;cerebellar loops that refine prefrontal control over emotional and self-referential processes (Schmahmann, 2019). Disruption of these loops has been proposed to contribute to psychopathological phenotypes characterized by impaired emotion regulation, impulsivity, and unstable self-representation. In adult BPD, cerebellar alterations have been associated with emotional dysregulation and impulsivity (Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Grecucci et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; 2025), raising the question of whether similar cerebellar involvement is already detectable during adolescence, when core symptoms first emerge. aBPD may involve early alterations in fronto\u0026ndash;cerebellar systems supporting the integration of emotional monitoring and cognitive control (Yi et al., 2026).\u003c/p\u003e \u003cp\u003eDespite these advances, the existing neuroimaging literature on aBPD is characterized by several methodological shortcomings. A primary limitation is the widespread reliance on mass-univariate analytical approaches, which treat each voxel as statistically independent and therefore fail to capture the coordinated nature of brain organization (Sorella et al., \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Lapomarda et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2021\u003c/span\u003e, 2022; Grecucci et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Most studies have employed voxel-based morphometry (VBM), a technique that may be insufficiently sensitive to subtle but spatially distributed patterns of structural alteration (Pappaianni et al., \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Furthermore, the generalizability of previous findings has been limited, reducing their potential translational value. Additional constraints include relatively small sample sizes (typically ranging from 13 to 53 participants), a strong overrepresentation of female patients (often exceeding 75\u0026ndash;100%), and incomplete control of key sociodemographic variables such as educational level. Finally, the restricted assessment of clinical and psychological dimensions in several studies has hampered a comprehensive interpretation of observed neural abnormalities. Together, these limitations underscore the need for large-sample, multivariate, and clinically well-characterized investigations capable of capturing distributed neurodevelopmental alterations in aBPD.\u003c/p\u003e \u003cp\u003eIn a recent large-scale neuroimaging study, Yi and colleagues (\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) investigated structural brain alterations in adolescents with borderline personality disorder using a multivariate, machine-learning\u0026ndash;based framework by combining source-based morphometry with supervised learning techniques. Their results revealed a distinctive pattern of gray matter covariance, characterized by increased volume within regions overlapping the posterior hub of the default mode network and the cerebellum, alongside reduced gray matter in frontal control regions.\u003c/p\u003e \u003cp\u003eFor what concerns white matter alterations, previous work in adolescents with borderline personality disorder reported reduced fractional anisotropy and increased mean diffusivity within cortical\u0026ndash;limbic and cortical\u0026ndash;thalamic pathways compared to healthy controls (Yi et al., \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), by using Diffusion tensor imaging, suggesting early alterations in WM organization. Disruptions in these WM properties have been proposed to impair large-scale network connectivity, with downstream consequences for emotion regulation and impulse control in BPD (Grottaroli et al., 2020; Kelleher-Unger et al., 2021). Consistent with this view, DTI studies and meta-analytic evidence in adults with BPD have identified WM abnormalities predominantly affecting the corpus callosum, corona radiata, external capsule, and uncinate fasciculus (Grottaroli et al., 2020; Kelleher-Unger et al., 2021). Increased RD within the corpus callosum has been linked to anxiety severity (Quattrini et al., 2019), while alterations in RD and AD have been reported across thalamic radiations, superior longitudinal fasciculus, and inferior fronto-occipital fasciculus (Gan et al., 2016; Maier-Hein et al., 2014; Ninomiya et al., 2018). In a recent diffusion MRI study, Yi and colleagues (\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) investigated white matter microstructural alterations associated with emotional dysfunction and childhood trauma in adolescents with borderline personality disorder. Using tract-based spatial statistics on diffusion tensor imaging data, the authors reported a pattern of altered radial and axial diffusivity within key cortico\u0026ndash;limbic pathways, including the corpus callosum, corona radiata, external capsule, and uncinate fasciculus. Specifically, adolescents with BPD showed increased radial diffusivity and reduced axial diffusivity in several callosal and fronto-limbic tracts compared to healthy controls. These findings provide evidence for early white matter abnormalities in adolescent BPD and suggest that WM microstructural measures may represent promising neurobiological markers linked to both clinical severity and adverse developmental experiences.\u003c/p\u003e \u003cp\u003eOne additional limitation of the previous study, is that they typically examined gray and white matter alterations separately, preventing a unified characterization of their joint contribution to emotion dysregulation and other symptoms. Converging evidence from adults with BPD points to a direction in which both structural properties of the brain are altered in these populations (Grecucci et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). As far as we know this methodology was never applied to adolescents with BPD. Building directly on our prior findings, the present study was therefore designed to extend this framework by integrating gray and white matter structural information within a unified unsupervised data-fusion approach, with the aim of identifying latent multimodal neurodevelopmental patterns underlying adolescent BPD.\u003c/p\u003e \u003cp\u003eIn recent years, machine learning (ML) has become increasingly central to neuroimaging data analysis, influencing not only methodological practice but also conceptual models of brain organization and disease (Nenning \u0026amp; Langs, 2022; Baggio et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Group-level inference methods are often poorly suited to capturing individual variability and distributed network-level alterations, both of which are essential for understanding complex cognitive and behavioral phenomena (Grecucci et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Hebart \u0026amp; Baker, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Vieira et al., \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Schnack, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Within this context, unsupervised learning techniques offer a powerful means of uncovering latent data structure through clustering and dimensionality reduction without relying on predefined outcome variables (Kherif \u0026amp; Latypova, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo test the presence of coupled and decoupled GM-WM networks in aBPD, we applied an unsupervised machine-learning data-fusion approach based on Transposed Independent Vector Analysis (tIVA) to identify independent neural circuits across modalities (Adali et al., 2015; Lee et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). tIVA was selected because it integrates key advantages of independent component analysis and canonical correlation analysis while overcoming central limitations of existing fusion frameworks. Unlike joint independent component analysis, which imposes a single mixing matrix across modalities and assumes equivalent signal contributions (Calhoun et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2006\u003c/span\u003e), tIVA estimates modality-specific mixing matrices while statistically coupling source vectors, allowing shared variance to emerge without enforcing identical spatial patterns. Compared with multiset-CCA or mCCA+jICA approaches (Sui et al., 2013), tIVA incorporates higher-order, non-Gaussian statistics, increasing sensitivity to subtle and potentially nonlinear cross-modal relationships. Benchmark studies have demonstrated that tIVA yields more stable components and improved diagnostic classification relative to joint-ICA or linked-ICA, while maintaining computational scalability as additional modalities are included (Adali et al., 2015). tIVA provides a principled framework for capturing distributed GM\u0026ndash;WM covariance patterns that are more functionally informative than traditional atlas-based representations (Biswal et al., 2010; Fox et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Baggio et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Grecucci et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWe expected to identify increased GM-WM concentration in regions belonging to the default mode network and to the cerebellum, together with reduced prefrontal control areas, consistent with prior functional and structural evidence in BPD (Koenigsberg et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e1999\u003c/span\u003e; Paris et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Ruocco et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). We expected that alterations would relate to specific facets of emotion regulation and executive functioning. We further hypothesized the presence of a multimodal gray matter\u0026ndash;white matter component reflecting coupled alterations within limbic\u0026ndash;affective circuits, alongside complementary changes in frontal and posterior default mode regions, in line with previous reports of prefrontal and white matter involvement in BPD (Hazlett et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Goethals et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Grecucci et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Finally, we predicted that the expression of these data-driven components would be meaningfully associated with core clinical features of adolescent BPD, particularly emotion dysregulation and self-injurious behaviors, and would contribute to distinguishing adolescents with BPD from healthy controls.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eParticipants\u003c/h2\u003e \u003cp\u003eAdolescents with borderline personality disorder (BPD) were consecutively recruited between October 2021 and October 2023 from the psychiatric clinics of the Mental Health Center at Xiangya Hospital, Central South University. Healthy control participants (HCs), matched for age, gender, and education level, were recruited from local schools during the same period.Inclusion criteria for the BPD group were: (1) age between 13 and 18 years; (2) fulfillment of DSM-IV diagnostic criteria for BPD (\u0026ge;\u0026thinsp;5 out of 9 criteria); (3) symptom stability for at least two years; and (4) a score\u0026thinsp;\u0026ge;\u0026thinsp;66 on the Borderline Personality Feature Scale for Children (BPFS-C). Diagnosis followed a rigorous two-step procedure. First, an associate professor of psychiatry with over 15 years of clinical experience applied DSM-5-TR criteria during routine clinical assessment. All participants meeting initial criteria subsequently completed the DSM-IV Axis II Structured Clinical Interview for Personality Disorders (SCID-II), administered by trained child and adolescent psychiatrists with at least five years of clinical experience. Final diagnoses were independently confirmed by two senior psychiatrists, ensuring adherence to internationally recognized diagnostic standards. To enhance internal validity in structural MRI analyses, exclusion criteria were applied for psychiatric and neurodevelopmental conditions known to produce strong, independent neuroanatomical alterations, including schizophrenia spectrum disorders, bipolar spectrum disorders, major depressive disorder, post-traumatic stress disorder, attention-deficit/hyperactivity disorder, substance use disorders, and neurological diseases. These exclusions were implemented to reduce neurobiological confounds rather than to imply a comorbidity-free presentation of adolescent BPD. Healthy controls underwent comprehensive screening, including medical history review and psychological assessment, to exclude any current or past psychiatric or neurological disorders. Additional exclusion criteria for all participants included IQ\u0026thinsp;\u0026le;\u0026thinsp;80, contraindications to MRI, substance use, and left-handedness, as assessed by the Edinburgh Handedness Inventory. The final sample comprised 129 adolescents with BPD (aged 12\u0026ndash;17) and 107 matched healthy controls. A priori power analysis indicated that this sample size provides 80% power to detect medium effect sizes (Cohen\u0026rsquo;s d\u0026thinsp;\u0026asymp;\u0026thinsp;0.5) at α\u0026thinsp;=\u0026thinsp;.05. Patients were recruited from a large tertiary psychiatric center receiving referrals from outpatient clinics, emergency services, community mental health providers, and inpatient units, ensuring a clinically representative sample of moderate-to-severe adolescent BPD. The study was approved by the Institutional Review Board of Xiangya Hospital, Central South University (IRB No. 2022020227). Written informed consent was obtained from the parents or legal guardians of all participants.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eClinical assessment\u003c/h3\u003e\n\u003cp\u003eDiagnostic assessment was conducted using standardized, structured interviews. The borderline personality disorder diagnosis was established using the BPD module of the Structured Clinical Interview for DSM-IV Personality Disorders (SCID-II), with diagnosis confirmed when at least five of the nine criteria were met. Axis I psychiatric disorders were assessed using the Schedule for Affective Disorders and Schizophrenia for School-Age Children\u0026mdash;Present and Lifetime Version (K-SADS-PL). Healthy control participants underwent the same structured interviews to exclude any personality or Axis I disorders. Interviews were conducted with both adolescents and their parents or legal guardians. Final diagnoses were determined following a comprehensive clinical evaluation by two experienced child and adolescent psychiatrists (QX and FJ, with 10 and 8 years of clinical experience, respectively). Information regarding psychiatric disorders in first-degree relatives was collected from participants and their parents or legal guardians. All clinical assessments were completed prior to MRI acquisition and on the same day. Intellectual functioning was evaluated using the Abbreviated Wechsler Intelligence Scale (Wechsler, \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e1999\u003c/span\u003e). Borderline personality traits were assessed using the Borderline Personality Features Scale for Children (BPFS-C), with a cutoff score of 66 indicating clinically significant borderline features. The BPFS-C includes four subscales assessing affective instability, identity problems, negative relationships, and self-harm. Emotion regulation difficulties were measured using the Difficulties in Emotion Regulation Scale (DERS; Li et al., \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), which comprises six subscales covering emotional awareness, emotional clarity, nonacceptance of emotional responses, impulse control difficulties, difficulties in goal-directed behavior, and limited access to emotion regulation strategies. Anxiety symptoms were evaluated using the Screen for Child Anxiety Related Emotional Disorders (SCARED; Birmaher et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e1999\u003c/span\u003e), which includes five subscales assessing panic/somatic symptoms, generalized anxiety, separation anxiety, social phobia, and school phobia. Global functioning was assessed with the clinician-rated Global Assessment of Functioning (GAF) scale, providing a composite index of psychological, social, and occupational functioning (Aas, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Depressive symptoms were measured using the Mood and Feelings Questionnaire (MFQ; Wood et al., \u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e1995\u003c/span\u003e), a 33-item self-report scale with higher scores indicating greater depressive severity.\u003c/p\u003e\n\u003ch3\u003eMRI data acquisition\u003c/h3\u003e\n\u003cp\u003eStructural MRI data were acquired on a Siemens MAGNETOM Prisma 3T scanner. For each participant, a high-resolution three-dimensional T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) sequence with whole-brain coverage was collected. Acquisition parameters were as follows: acquisition matrix 256 \u0026times; 256 mm\u0026sup2;, isotropic voxel size 10 mm\u0026sup3;, repetition time (TR)\u0026thinsp;=\u0026thinsp;2300 ms, echo time (TE)\u0026thinsp;=\u0026thinsp;203 ms, flip angle\u0026thinsp;=\u0026thinsp;9\u0026deg;, and 176 contiguous sagittal slices. All images were visually inspected by a senior neuroradiologist (ZH; \u0026gt;30 years of experience) to ensure adequate image quality and to exclude incidental brain abnormalities.\u003c/p\u003e\n\u003ch3\u003ePreprocessing\u003c/h3\u003e\n\u003cp\u003eAfter initial quality control to exclude images with artifacts, all MRI data were preprocessed using a unified pipeline implemented in the Computational Anatomy Toolbox (CAT12; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.neuro.uni-jena.de/cat/\u003c/span\u003e\u003cspan address=\"http://www.neuro.uni-jena.de/cat/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) running within SPM12 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.fil.ion.ucl.ac.uk/spm/\u003c/span\u003e\u003cspan address=\"http://www.fil.ion.ucl.ac.uk/spm/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and MATLAB. Images were segmented into gray matter, white matter, and cerebrospinal fluid, and modulated normalized maps were generated. High-dimensional spatial registration was performed using Diffeomorphic Anatomical Registration through Exponential Lie Algebra (DARTEL), followed by normalization to Montreal Neurological Institute (MNI) space. Finally, the normalized images were spatially smoothed using an 8-mm full-width at half-maximum Gaussian kernel.\u003c/p\u003e\n\u003ch3\u003eTransposed Independent Vector Analysis\u003c/h3\u003e\n\u003cp\u003eTo identify distributed neural circuits underlying adolescent BPD, we employed an unsupervised machine-learning approach based on Transposed Independent Vector Analysis (tIVA) (Adali et al., 2015; Lee et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). tIVA is a multivariate source separation technique specifically designed for multimodal data fusion, allowing the joint decomposition of multiple imaging modalities\u0026mdash;here gray matter (GM) and white matter (WM)\u0026mdash;into statistically independent components while preserving modality-specific information. In contrast to conventional univariate or concatenation-based approaches, tIVA estimates separate mixing matrices for each modality while statistically coupling the corresponding source vectors, thereby enabling the identification of latent GM\u0026ndash;WM covariance patterns without imposing identical spatial distributions across modalities. The tIVA analysis was implemented using the Fusion ICA Toolbox (FIT; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://mialab.mrn.org/software/fit\u003c/span\u003e\u003cspan address=\"http://mialab.mrn.org/software/fit\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (Calhoun et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2006\u003c/span\u003e) within the MATLAB 2018a environment (The MathWorks, Natick, MA, USA). Preprocessed GM and WM maps were entered simultaneously into the model. Prior to decomposition, the optimal number of independent vectors was estimated using the minimum description length (MDL) criterion, which provides a data-driven balance between model complexity and explanatory power. Independent Vector Analysis (IVA) extends Independent Component Analysis (ICA) to the joint analysis of multiple related datasets by explicitly modeling statistical dependence across datasets, in addition to enforcing independence within each dataset (Adali et al., 2015; Lee et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). In this framework, each dataset \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}^{\\left(k\\right)}\\)\u003c/span\u003e\u003c/span\u003e(with \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k=1,\\dots\\:,K\\)\u003c/span\u003e\u003c/span\u003e) is assumed to arise from a linear mixture of latent sources \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{S}^{\\left(k\\right)}\\)\u003c/span\u003e\u003c/span\u003ethrough modality-specific mixing matrices \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{A}^{\\left(k\\right)}\\)\u003c/span\u003e\u003c/span\u003e. Source estimates are obtained via corresponding demixing matrices \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{W}^{\\left(k\\right)}\\)\u003c/span\u003e\u003c/span\u003e, yielding subject- or voxel-wise source estimates \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{u}^{\\left(k\\right)}\\)\u003c/span\u003e\u003c/span\u003e. A defining feature of IVA is the introduction of the source component vector (SCV), which is constructed by grouping corresponding sources across the Kdatasets. Each SCV therefore represents a multivariate entity that captures statistical dependence across modalities for a given component, while different SCVs are assumed to be mutually independent. This formulation allows IVA to exploit cross-dataset correlations when they are present, while reducing to standard ICA when a single dataset is analyzed (K=1). The IVA objective function is defined in terms of the mutual information rate across SCVs. Minimization of this criterion simultaneously promotes independence between different SCVs and dependence within the elements of each SCV. As a result, IVA can leverage multiple forms of statistical diversity, including higher-order statistics and sample dependence, thereby improving source identifiability and separation performance relative to ICA. Importantly, this property allows IVA to identify sources that may be Gaussian within individual datasets but become identifiable due to correlations across datasets\u0026mdash;an advantage that is particularly relevant in multimodal fusion settings. Transposed Independent Vector Analysis (tIVA) further generalizes this framework by applying IVA to transposed data representations, such that the independent sources correspond to subject-wise expression profiles rather than spatial maps. In this formulation, tIVA effectively extends multi-set canonical correlation analysis (MCCA) by relaxing orthogonality constraints and incorporating higher-order statistical information. The tIVA output consists of subject-specific loading coefficients and spatial maps for each modality. Spatial maps were transformed into Talairach space to facilitate anatomical labeling and interpretation of the associated brain regions. Both positive and negative voxel weights were retained, reflecting regions contributing in opposite directions within each component. For visualization, component maps were rendered using Surf Ice (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nitrc.org/projects/surfice/\u003c/span\u003e\u003cspan address=\"https://www.nitrc.org/projects/surfice/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (Rorden), allowing intuitive inspection of the spatial extent and anatomical distribution of the identified networks. Group-level statistical analyses were performed on the component loading coefficients using independent-samples t-tests implemented in JASP (version 0.16.3; JASP Team, 2023) to assess differences between adolescents with BPD and healthy controls. This analytic strategy enabled us to quantify how specific multimodal network components differentiated diagnostic groups and to characterize their contribution to the structural phenotype of adolescent BPD.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePredictive model\u003c/h2\u003e \u003cp\u003eSubject-specific loading coefficients derived from tIVA decomposition were subsequently entered into a supervised machine learning framework to classify adolescents with borderline personality disorder (aBPD) versus healthy controls (HCs). Supervised classification was implemented using a decision tree classifier as provided in JASP [65]. In this framework, diagnostic group membership (aBPD vs HC) was modeled as a function of the gray and white matter networks loadings. The decision tree algorithm recursively partitions the predictor space by selecting, at each step, the network component and corresponding threshold that best separates participants into subgroups with more homogeneous class labels. For each candidate split, the algorithm evaluates classification impurity\u0026mdash;defined as the degree of class mixing within each node\u0026mdash;using standard criteria such as misclassification error or Gini impurity. The split that maximally reduces impurity, that is, that produces the greatest improvement in class separation between aBPD and HC participants, is selected. As a result, each terminal node of the tree represents a subgroup of participants characterized by a specific pattern of network expression and a predominant diagnostic label. The predicted class for a new participant corresponds to the majority class within the terminal node to which that participant is assigned. Model generalizability was assessed using a hold-out validation strategy, with 80% of the observations used to train the classifier and the remaining 20% reserved for testing its performance on previously unseen cases. This procedure ensured that classification accuracy reflected true out-of-sample performance. Model hyperparameters were set as follows: a minimum of 20 observations was required to attempt a split, terminal nodes were constrained to contain at least 7 observations, maximum tree depth was set to 30, and the complexity parameter was fixed at 0.01 to penalize overly complex trees. All predictors were standardized prior to model estimation.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eBehavioral findings\u003c/h2\u003e \u003cp\u003eDemographic and clinical characteristics of adolescents with borderline personality disorder and healthy controls are reported in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The two groups did not differ significantly in age (Mann\u0026ndash;Whitney U\u0026thinsp;=\u0026thinsp;7621, p\u0026thinsp;=\u0026thinsp;0.154), years of education (U\u0026thinsp;=\u0026thinsp;6200, p\u0026thinsp;=\u0026thinsp;0.163), or gender distribution (χ\u0026sup2; = 3.217, p\u0026thinsp;=\u0026thinsp;0.073). In contrast, adolescents with BPD showed markedly elevated symptom severity across all clinical measures, including borderline personality features (BPFS), depressive symptoms (MFQ), anxiety symptoms (SCARED), and difficulties in emotion regulation (DERS), alongside significantly lower levels of global functioning as indexed by GAF scores, relative to healthy controls (all p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eBehavioral and clinical findings\u003c/h2\u003e \u003cp\u003eAdolescents with borderline personality disorder (aBPD) and healthy controls (HCs) did not differ significantly in basic demographic variables, including gender distribution (aBPD: 42 males, 87 females; HC: 47 males, 60 females; p\u0026thinsp;=\u0026thinsp;0.073), age (mean age: aBPD\u0026thinsp;=\u0026thinsp;15.55 years; HC\u0026thinsp;=\u0026thinsp;15.90 years; p\u0026thinsp;=\u0026thinsp;0.154), or years of education (aBPD\u0026thinsp;=\u0026thinsp;9.82 years; HC\u0026thinsp;=\u0026thinsp;9.65 years; p\u0026thinsp;=\u0026thinsp;0.163). The mean duration of illness in the aBPD group was 12.85 years.In contrast, pronounced differences emerged across all clinical and psychological measures. Adolescents with aBPD showed markedly elevated borderline personality features, as indexed by the Borderline Personality Features Scale for Children (BPFS-C). Specifically, the total BPFS-C score was substantially higher in the aBPD group compared to HCs (86.67 vs. 56.00, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This pattern was consistent across all BPFS-C subscales, including affective instability (BPFS-C-1: 23.50 vs. 14.16, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), identity problems (BPFS-C-2: 21.84 vs. 14.71, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), negative relationships (BPFS-C-3: 20.76 vs. 14.29, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and self-harm (BPFS-C-4: 20.75 vs. 12.82, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).Adolescents with aBPD also exhibited significantly greater difficulties in emotion regulation, as measured by the Difficulties in Emotion Regulation Scale (DERS). The total DERS score was markedly higher in the aBPD group compared to HCs (123.90 vs. 80.32, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). All DERS subdomains showed significant group differences, including emotional awareness (DERS-A: 19.41 vs. 15.74, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), emotional clarity (DERS-B: 15.32 vs. 10.41, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), nonacceptance of emotional responses (DERS-C: 18.38 vs. 11.95, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), impulse control difficulties (DERS-D: 21.20 vs. 11.49, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), difficulties engaging in goal-directed behavior (DERS-E: 20.62 vs. 12.68, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and limited access to emotion regulation strategies (DERS-F: 29.89 vs. 17.04, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).Anxiety-related symptoms were also significantly elevated in adolescents with aBPD, as assessed by the Screen for Child Anxiety Related Emotional Disorders (SCARED). The total SCARED score was more than doubled in the aBPD group relative to HCs (46.35 vs. 19.16, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This difference was observed across all examined anxiety domains, including panic/somatic symptoms (SCARED-A: 15.00 vs. 4.48, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), generalized anxiety (SCARED-B: 11.51 vs. 4.98, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and separation anxiety (SCARED-C: 6.50 vs. 2.83, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Overall, these results indicate that adolescents with aBPD present a pervasive and multidimensional clinical profile characterized by severe borderline personality features, marked emotion regulation deficits, and elevated anxiety symptomatology, despite a demographic profile comparable to that of healthy peers.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eNeural findings\u003c/h2\u003e \u003cp\u003eApplication of transposed Independent Vector Analysis (tIVA) to gray and white matter structural MRI data identified 20 independent vectors (tIVs) reflecting covarying neuroanatomical patterns across subjects. Group differences between adolescents with borderline personality disorder (aBPD) and healthy controls (HCs) were assessed on the subject-specific loading coefficients using independent-samples t-tests. To account for multiple comparisons, a Bonferroni-corrected significance threshold was applied. At this corrected threshold, two components showed significant group differences, corresponding to a gray matter\u0026ndash;only frontal\u0026ndash;cerebellar network (tIV18) and a coupled gray\u0026ndash;white matter affective network (tIV05). tIV18 showed a robust group difference in gray matter loadings (t\u0026thinsp;=\u0026thinsp;4.131, df\u0026thinsp;=\u0026thinsp;231, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In contrast, no significant group difference was observed for the white matter loadings of this component (t\u0026thinsp;=\u0026thinsp;0.136, df\u0026thinsp;=\u0026thinsp;231, p\u0026thinsp;=\u0026thinsp;0.892), indicating a decoupled gray matter\u0026ndash;only effect. tIV18 was characterized by increased\u003c/p\u003e \u003cp\u003ereduced GMC in adolescents with aBPD within medial fronto-temporo-parietal regions and the cerebellum, as well as decreased lateral fronto-parietal GMC. tIV05 showed significant group differences in both imaging modalities, consistent with a multimodal gray\u0026ndash;white matter coupled pattern. Gray matter loadings differed significantly between groups (t\u0026thinsp;=\u0026thinsp;3.488, df\u0026thinsp;=\u0026thinsp;231, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Importantly, a significant group difference was also observed for the corresponding white matter loadings (t\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;3.165, df\u0026thinsp;=\u0026thinsp;231, p\u0026thinsp;=\u0026thinsp;0.002). Spatially, tIV05 was characterized by increased gray matter concentration in adolescents with aBPD within temporal regions, anterior cingulate cortex, and temporal poles, including the amygdala and hippocampus. Moreover, tIV05 showed also a reduction in frontal regions and in the precuneus. For what concerns the WM counterpart, WM portions adjacent to temporo-polar, parietal and cerebellar regions showed a decreased, whereas regions adjacent to the basal ganglia, precuneus and frontal regions displayed a reduction coherently.\u003c/p\u003e \u003cp\u003eNone of the significant tIV components differed as a function of gender, nor were their loading coefficients significantly correlated with age or years of education (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.1), indicating that the observed group effects were not driven by demographic variables. See Tables\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003eA-B, \u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e2\u003c/span\u003eA-B, \u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e3\u003c/span\u003eA-B, and Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003etIV05 was characterized by increased gray matter concentration in adolescents with aBPD within temporal regions, anterior cingulate cortex, and temporal poles and by a reduction in frontal regions and in the precuneus.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003etIV displayed also an increase in GMC in the amygdala, here 3D reconstructed (top). For what concerns the WM counterpart (bottom), WM portions adjacent to temporo-polar, parietal and cerebellar regions showed a decreased, whereas regions adjacent to the basal ganglia, precuneus and frontal regions displayed a reduction coherently.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eA \u0026ndash; tIV18\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV18 \u0026ndash; increased GMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10.2/10.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.8 (-43, -68, -18)/8.8 (24, -82, -15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLingual Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17, 18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.8/2.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.6 (-15, -85, -13)/8.5 (19, -84, -13)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFusiform Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18, 19, 20, 37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.6/2.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.6 (-25, -87, -16)/8.3 (21, -79, -14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUvula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.3/2.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.8 (-30, -64, -23)/6.8 (33, -80, -23)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCulmen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7.0/5.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.3 (-46, -51, -28)/5.7 (43, -52, -24)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCerebellar Tonsil\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.8/2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.2 (-49, -51, -32)/5.3 (49, -54, -34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePyramis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.0/2.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.4 (-48, -64, -32)/5.0 (34, -74, -33)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Semi-Lunar Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5/3.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.0 (-45, -66, -35)/4.9 (22, -76, -38)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17, 18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.1 (-24, -88, -12)/5.0 (37, -81, -14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-Gyral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.8 (-3, -81, -9)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.7 (24, -86, -10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.7 (0, -71, 46)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclive of Vermis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.6 (3, -72, -12)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.5 (4, 7, 55)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedial Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.5 (1, 41, 39)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eB \u0026ndash; tIV18\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV18 \u0026ndash; Decreased GMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-Gyral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8 (-55, -48, -5)/6.5 (33, -72, 26)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.9 (42, 6, 37)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.8 (31, -78, 20)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.4 (33, -76, 27)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.2 (28, -75, 20)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8, 9, 10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.9 (-21, 62, 8)/5.2 (39, 8, 40)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6, 9, 46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.7 (-52, 31, 3)/4.9 (49, 6, 33)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21, 37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.1/0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-58, -41, -8)/4.8 (31, -72, 20)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.7 (21, -68, 10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Cingulate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.4 (24, -65, 9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Parietal Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.6/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2 (-50, -37, 24)/4.4 (52, -29, 24)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInsula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13, 40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.6 (-48, -37, 21)/4.2 (50, -25, 18)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostcentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.1 (53, -27, 21)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20, 37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.1 (-46, -2, -34)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLateral Ventricle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.9 (27, -62, 9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSupramarginal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.8 (50, -47, 27)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.5 (-28, 54, -3)/3.7 (24, 58, -8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAngular Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.6 (36, -76, 30)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.5 (46, -27, 15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eA - tIV05\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV05 \u0026ndash; Increased GMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAmygdala\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20, 28, 36, 38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.8/3.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.0 (-24, 4, -34)/8.3 (24, 0, -34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22, 38, 42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.7/3.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.6 (-25, 9, -33)/7.6 (24, 9, -32)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCerebellar Tonsil\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.6/1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.4 (-12, -55, -41)/5.5 (15, -56, -41)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-Gyral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4/0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.8 (-28, 3, -35)/7.2 (27, 3, -34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20, 21, 38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.7/0.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.0 (-42, 12, -32)/5.9 (33, 8, -37)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Semi-Lunar Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.0/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.8 (-3, -57, -41)/5.0 (4, -60, -41)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.1/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.5 (-31, -3, -39)/4.2 (31, -8, -36)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Parietal Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/3.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/6.1 (61, -40, 24)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThalamus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.4/2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.6 (-9, -13, 8)/5.8 (10, -11, 9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSupramarginal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.6 (59, -42, 33)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.3/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.2 (-10, -59, 7)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParahippocampal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19, 30, 35, 36, 37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.4/1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.9 (-31, -17, -24)/5.2 (33, -17, -26)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCulmen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.8/2.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.5 (-6, -61, -3)/5.0 (19, -51, -10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUvula of Vermis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.0 (-1, -61, -31)/4.1 (3, -61, -34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Cingulate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.4/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.9 (-10, -59, 11)/3.7 (19, -51, 7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLingual Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18, 19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.8 (-10, -57, 4)/3.8 (1, -86, -6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCulmen of Vermis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-1, -63, -4)/4.4 (3, -63, -6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.8/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.5 (-19, -56, -11)/4.7 (19, -53, -14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtra-Nuclear\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.3/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.5 (-21, -54, 7)/4.1 (9, -4, 4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostcentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2, 40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.5 (-53, -20, 33)/3.7 (55, -34, 49)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFusiform Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19, 20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.6/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (-21, -56, -7)/3.7 (27, -36, -16)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUvula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2 (-4, -64, -31)/4.0 (1, -61, -28)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eB - tIV05\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV05 \u0026ndash; Decreased GMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Semi-Lunar Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.7/1.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.2 (-30, -75, -39)/5.8 (24, -72, -42)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18, 19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/5.5 (30, -85, 10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePyramis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.2/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-19, -82, -34)/3.6 (37, -69, -33)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10, 45, 46, 47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.5/1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2 (-49, 27, -5)/4.4 (50, 24, -6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUvula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (-10, -80, -34)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCingulate Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.0/0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (0, -42, 30)/4.1 (3, -39, 31)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (-40, 35, 22)/3.7 (33, 38, 23)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.6/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (-9, -46, 33)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.9 (34, -75, -6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCerebellar Tonsil\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.9 (-40, -62, -41)/3.9 (42, -60, -42)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostcentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8 (-55, -19, 49)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eA \u0026ndash; tIV05\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV05 \u0026ndash; Increased WMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLentiform Nucleus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.9/3.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.8 (-12, 0, 4)/9.6 (12, 0, 1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtra-Nuclear\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5/7.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.2 (-3, -36, 14)/9.0 (13, -1, 6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThalamus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.9 (-9, -6, 6)/7.6 (10, -4, 6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9, 13, 44, 45, 46, 47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.6/4.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.3 (-42, 28, 8)/7.6 (49, 27, 14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-Gyral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.3/4.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.6 (-42, 15, 19)/6.8 (46, 24, 14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/6.3 (18, -94, 12)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8, 9, 11, 46, 47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.5/0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.2 (-30, 31, 32)/5.4 (46, 30, 14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Occipital Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/1.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2 (-24, -88, 18)/5.7 (33, -78, 14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.4 (-31, 35, 31)/3.6 (16, 14, 50)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCaudate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.7 (9, 6, 4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostcentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.6 (-52, -20, 34)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInsula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.9 (-39, 18, 12)/4.5 (40, 18, 12)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.3/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.4 (-43, -60, 18)/3.5 (58, -47, -3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Parietal Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.2 (34, -62, 43)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Cingulate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.3/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.0 (-3, -32, 22)/3.6 (4, -28, 24)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.0 (-13, -62, -15)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLateral Ventricle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8 (-12, -36, 17)/3.9 (15, -39, 14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eB \u0026ndash; tIV05\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003etIV05 \u0026ndash; Decreased WMC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBrodmann Area\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003evolume (cc)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003erandom effects: Max Value (x, y, z)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Frontal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/1.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.9 (-42, 32, 22)/8.0 (42, 28, 26)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-Gyral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.3/0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.8 (-39, -9, -26)/6.7 (39, 27, 24)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.5/0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.1 (-37, -2, -33)/4.1 (40, 1, -30)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAngular Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.8 (-48, -58, 35)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Parietal Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.8/1.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-48, -58, 39)/5.7 (43, -52, 45)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.6/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.7 (-34, -2, -35)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeclive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.9/1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.3 (-19, -78, -20)/5.6 (13, -78, -19)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFusiform Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.3/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.2 (-42, -3, -26)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSupramarginal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4/0.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.1 (-46, -55, 32)/4.8 (42, -48, 34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7, 19, 31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.6/0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-19, -80, 34)/5.0 (24, -74, 35)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePyramis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.7/0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-10, -80, -26)/4.8 (10, -78, -25)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuperior Temporal Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21, 38, 41, 42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4/1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3 (-56, -13, 1)/4.8 (59, -9, -1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUvula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.2/0.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-34, -71, -23)/4.6 (13, -80, -23)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCuneus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7, 18, 19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.8/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-9, -75, 24)/4.3 (22, -74, 31)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTuber\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.2/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7 (-39, -68, -23)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUncus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.5 (-36, -9, -30)/3.8 (33, -5, -34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecentral Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/4.4 (39, 13, 34)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCerebellar Tonsil\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.5 (-40, -60, -41)/3.9 (36, -59, -42)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtra-Nuclear\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8 (-31, -10, 6)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInferior Semi-Lunar Lobule\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.2/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.7 (-40, -63, -38)/3.6 (25, -66, -42)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLingual Gyrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.7 (-10, -80, -6)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInsula\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.0/0.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-999.0 (0, 0, 0)/3.7 (42, -37, 18)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCulmen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.1/0.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.5 (-40, -54, -29)/-999.0 (0, 0, 0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eBrain \u0026ndash; symptoms correlations\u003c/h2\u003e \u003cp\u003eCorrelational analyses were conducted between the loading coefficients of the significant tIVA components (tIV05 and tIV18) and the clinical questionnaires, controlling for gender, age, and years of education.\u003c/p\u003e \u003cp\u003eComponent tIV18 was significantly associated with the Nonacceptance of Emotional Responses subscale (DERS-C; Spearman\u0026rsquo;s ρ\u0026thinsp;=\u0026thinsp;0.207, p\u0026thinsp;=\u0026thinsp;0.020). tIV18 did not show significant associations with the SCARED total score (Spearman\u0026rsquo;s ρ = \u0026minus;0.095, p\u0026thinsp;=\u0026thinsp;0.292) or with any of the SCARED subscales (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.08). Component tIV05 showed a significant negative association with the Emotional Awareness subscale of the DERS (DERS-A; Spearman\u0026rsquo;s ρ = \u0026minus;0.241, p\u0026thinsp;=\u0026thinsp;0.007), indicating that higher expression of this network was related to greater difficulties in emotional awareness. tIV18 showed a significant association with Trail Making Test part A (TMT-A) (Spearman\u0026rsquo;s ρ = \u0026minus;0.233, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating that higher expression of this component was associated with poorer processing speed and attentional performance. In addition, tIV18 was significantly negatively correlated with duration of illness (Spearman\u0026rsquo;s ρ = \u0026minus;0.628, p\u0026thinsp;=\u0026thinsp;0.019).\u003c/p\u003e \u003cp\u003eNo significant correlations were observed between tIV18 and remaining questionnaires (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003eComponent tIV05 showed a significant negative association with overall anxiety severity, as measured by the SCARED total score (Spearman\u0026rsquo;s ρ = \u0026minus;0.198, p\u0026thinsp;=\u0026thinsp;0.026). In addition, tIV05 was significantly negatively correlated with the School Phobia subscale of the SCARED (SCARED-D; Spearman\u0026rsquo;s ρ = \u0026minus;0.198, p\u0026thinsp;=\u0026thinsp;0.026). tIV05 showed a significant negative association with the SSI-A subscale (Spearman\u0026rsquo;s ρ = \u0026minus;0.223, p\u0026thinsp;=\u0026thinsp;0.012), indicating that higher expression of this component was associated with greater severity of self-injurious attitudes. In addition, tIV05 was strongly negatively correlated with SSI-5 (Spearman\u0026rsquo;s ρ = \u0026minus;0.392, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Component tIV05 showed a significant negative association with the A-DES-II-A subscale (Spearman\u0026rsquo;s ρ = \u0026minus;0.323, p\u0026thinsp;=\u0026thinsp;0.035), indicating that higher expression of this component was related to greater dissociative symptoms in this domain. No significant correlations were observed between tIV05 and the A-DES-II total score (Spearman\u0026rsquo;s ρ = \u0026minus;0.294, p\u0026thinsp;=\u0026thinsp;0.056). tIV05 showed a significant positive association with global functioning, as assessed by the Global Assessment of Functioning (GAF) scale (Spearman\u0026rsquo;s ρ\u0026thinsp;=\u0026thinsp;0.257, p\u0026thinsp;=\u0026thinsp;0.004), indicating that higher expression of this component was related to better overall psychological, social, and occupational functioning. tIV05 showed a significant negative association with PANSS item 5 (Spearman\u0026rsquo;s ρ = \u0026minus;0.222, p\u0026thinsp;=\u0026thinsp;0.037), indicating that higher expression of this component was related to lower severity of this psychopathological dimension. In addition, tIV05 was significantly negatively correlated with Stroop condition 1 performance (Spearman\u0026rsquo;s ρ = \u0026minus;0.196, p\u0026thinsp;=\u0026thinsp;0.028). No significant correlations were observed between tIV05 and remaining questionnaires (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eDecision tree\u003c/h2\u003e \u003cp\u003eTo assess the predictive value of the identified structural networks at the individual level, a supervised decision tree classification was implemented to discriminate adolescents with BPD from healthy controls. The model was trained on 149 participants, validated on 38, and tested on an independent hold-out sample of 46 subjects, yielding a test accuracy of 71.7%, with comparable performance across classes. The model achieved an area under the ROC curve (AUC) of 0.687 (see Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), indicating moderate but reliable discriminative ability. Classification performance was higher for the BPD class, with a precision of 0.80, recall of 0.774, and F1 score of 0.787, compared to controls (precision\u0026thinsp;=\u0026thinsp;0.563; recall\u0026thinsp;=\u0026thinsp;0.600; F1\u0026thinsp;=\u0026thinsp;0.581). The Matthews correlation coefficient (MCC\u0026thinsp;=\u0026thinsp;0.368) further supported above-chance classification while accounting for class imbalance. Feature importance analysis highlighted the gray matter component tIV18-GM as the most influential predictor, showing the highest relative importance (20.8%) and the largest mean dropout loss, followed by white matter components corresponding to frontal and fronto-limbic pathways (tIV05-WM and tIV05-WM), thus confirming t-tests results but also extending to other components. Notably, the gray matter affective component tIV05 also contributed to classification, albeit with lower relative importance, suggesting a complementary role in refining group separation. The structure of the decision tree (see Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), further underscored the central role of the medial fronto-parietal\u0026ndash;cerebellar network (tIV18), which constituted the root node of the model. Initial splits based on tIV18 loadings effectively separated a large proportion of BPD and control participants, with subsequent refinement driven by white matter components, particularly those involving frontal and temporo-limbic connections. This hierarchical organization indicates that gray matter alterations provide the primary axis of discrimination, while white matter features modulate classification at subsequent levels, consistent with a network-based model in which macrostructural alterations constrain, but do not fully determine, diagnostic status. Overall, these results demonstrate that a limited set of data-driven structural brain networks\u0026mdash;especially the frontal\u0026ndash;cerebellar gray matter component and its associated white matter pathways\u0026mdash;can support individual-level classification of adolescent BPD with clinically meaningful accuracy. Importantly, the convergence between unsupervised (tIVA) and supervised (decision tree) analyses reinforces the robustness and translational relevance of the identified networks as potential neurobiological markers of borderline pathology in adolescence.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe present study extends our previous work on structural brain alterations in adolescents with borderline personality disorder (aBPD) by adopting a more advanced, multimodal data-fusion approach that jointly models gray and white matter covariance patterns. Whereas our earlier study relied on source-based morphometry applied exclusively to gray matter, the current investigation leverages transposed Independent Vector Analysis (tIVA) to disentangle both modality-specific and coupled GM\u0026ndash;WM networks. This allowed us to refine and expand previous findings, providing a more comprehensive neurobiological account of emotional and behavioral dysregulation in adolescent BPD. Consistent with our prior hypotheses and results, we identified two large-scale neuroanatomical systems that reliably differentiated aBPD from healthy controls: a medial fronto-parietal\u0026ndash;cerebellar gray matter network (tIV18) and a multimodal affective GM\u0026ndash;WM network (tIV05). Importantly, these networks showed dissociable patterns of clinical association, suggesting that distinct structural systems underlie different dimensions of borderline psychopathology during adolescence. In the next section we discuss the results in detail.\u003c/p\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eMedial fronto-parietal\u0026ndash;cerebellar alterations and deficits in emotion regulation and cognitive control (tIV18)\u003c/h2\u003e \u003cp\u003eIn our previous SBM study, one of the most robust findings was the involvement of the cerebellum and posterior default mode network (DMN) hubs, which showed increased gray matter concentration in adolescents with BPD and were associated with difficulties in emotion regulation, self-harm, anxiety symptoms, and reduced global functioning. These results converged with a growing body of evidence indicating that the cerebellum plays a critical role in emotional regulation, anxiety, depression, and self-injurious behaviors, beyond its classical motor functions (Salvador et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Koziol et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Hallett et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). In adult BPD, posterior cerebellar abnormalities have similarly been linked to impulsivity and emotional dysregulation, and cerebellar\u0026ndash;limbic interactions have been proposed as mechanisms underlying maladaptive coping strategies such as non-suicidal self-injury (Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Koziol et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Dadomo et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The present findings partially converge with, but also refine, this interpretation. The tIV18 component encompasses medial prefrontal, posterior midline, and cerebellar regions that together map onto a medial fronto\u0026ndash;cerebellar loop supporting self-referential monitoring, emotional appraisal, and cognitive efficiency. Within contemporary models of cerebellar function, such loops are thought to provide predictive modulation and error signaling to prefrontal systems, thereby constraining emotional responses and internal evaluative processes. Crucially, tIV18 exhibited a gray matter\u0026ndash;only pattern, with no corresponding white matter alterations, suggesting that during adolescence structural deviations within medial fronto-parietal\u0026ndash;cerebellar systems may primarily reflect regional morphological alterations rather than disrupted long-range connectivity.\u003c/p\u003e \u003cp\u003eFunctionally, tIV18 was selectively associated with nonacceptance of emotional responses, a core dimension of emotion dysregulation in BPD. This association is consistent with the role of medial prefrontal and posterior midline regions, key hubs of the default mode network (DMN), in self-referential and evaluative processing, including internal dialogue and affective appraisal (Menon, 2011; Grecucci et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Alterations within these systems have been linked to distorted self-related processing and maladaptive internal evaluation, hallmark features of borderline pathology, and DMN hyperactivity at rest has been repeatedly associated with difficulties in affective and interpersonal regulation (Koenigsberg et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e1999\u003c/span\u003e; Ruocco et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). From a structural perspective, integrity of posterior DMN regions has also been related to the capacity to engage adaptive emotion regulation strategies such as cognitive reappraisal (Ghoumroudi et al., 2023). Compared to our previous study, the clinical association profile of tIV18 appears more circumscribed: rather than correlating broadly with anxiety and self-harm, this network was specifically linked to emotional nonacceptance, attentional efficiency (as indexed by TMT-A), and illness duration. This pattern suggests that medial fronto-parietal\u0026ndash;cerebellar alterations may index a more fundamental vulnerability related to emotional monitoring and cognitive efficiency, potentially shaping illness course rather than momentary symptom severity.\u003c/p\u003e \u003cp\u003eTaken together, we put forward the hypothesis that adolescent BPD reflects a dysfunction of fronto\u0026ndash;cerebellar loops that normally orchestrate adaptive monitoring and modulation of internal emotional states, rather than being reducible to limbic hyperreactivity alone. We propose that early structural deviations within these circuits disrupt the predictive calibration of emotional responses and the tolerance of aversive affect, thereby fostering persistent emotional nonacceptance and a rigid, maladaptive style of self-regulation that is central to borderline pathology.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eAffective GM\u0026ndash;WM network alterations and emotional\u0026ndash;behavioral dysregulation (tIV05)\u003c/h2\u003e \u003cp\u003eThe second component, tIV05, represents a key methodological and clinical advance by demonstrating that anxiety, self-injurious behaviors, and functional impairment in adolescent BPD are more coherently captured by a coupled gray\u0026ndash;white matter affective network rather than by gray matter alterations alone. This network encompasses temporal regions, anterior cingulate cortex, temporal poles, and limbic structures such as the amygdala and hippocampus, together with convergent white matter alterations in frontal, parietal, and temporo-limbic pathways. Importantly, this configuration closely maps onto affective\u0026ndash;limbic circuits central to emotional reactivity and affect regulation in BPD. Functional neuroimaging studies have consistently reported exaggerated limbic responses to negative emotional stimuli in BPD, including heightened activation of the amygdala, hippocampus, parahippocampal regions, superior frontal gyrus, and cerebellum (LeBoeuf et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), while structural and functional abnormalities within these circuits have been linked to anxiety, heightened emotional sensitivity, and self-injurious behaviors across adolescent and adult samples (Schulze et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Quattrini et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Langerbeck et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The present findings extend this literature by showing that such affective disturbances are supported by network-level GM\u0026ndash;WM dyscoordination rather than by isolated regional alterations. Clinically, tIV05 was associated with greater anxiety severity, school-related anxiety, self-injurious attitudes and behaviors, and dissociative symptoms, while also showing a positive association with global functioning (GAF). This pattern suggests that tIV05 indexes a core affective liability in adolescent BPD that directly relates to clinically relevant dimensions of emotional and behavioral dysregulation, including self-harm risk. At the same time, its association with global functioning indicates that heightened affective sensitivity may coexist with relatively preserved functioning depending on contextual and regulatory factors, consistent with the marked heterogeneity observed in adolescent BPD. Crucially, the involvement of white matter within tIV05 highlights disrupted communication between limbic emotional generators and frontal regulatory systems as a central neurobiological substrate of affective dysregulation in aBPD. This finding strengthens fronto-limbic disconnection models of BPD by showing that anxiety, dissociation, and self-harm are more tightly linked to impaired structural integration than to gray matter abnormalities alone, in line with adult evidence of altered neurodevelopmental and myelination trajectories in affective and executive networks (Quattrini et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Overall, tIV05 emerges as a clinically meaningful affective network closely tied to the dimensions that most strongly drive clinical severity, risk, and treatment need in adolescent BPD.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eA model for aBPD and its implications\u003c/h2\u003e \u003cp\u003eIn our previous study, reduced gray matter concentration in lateral prefrontal regions,corresponding to the central executive network (CEN), was interpreted as reflecting deficits in top-down control and impulse regulation in adolescent BPD. This interpretation was supported by converging structural and functional evidence of dorsolateral prefrontal cortex abnormalities in both adolescents and adults with BPD (Hazlett et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Brunner et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Bozzatello et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), as well as functional hypometabolism in prefrontal regions implicated in cognitive control (Goethals et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Soloff et al., \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). The present findings refine this view by showing that prefrontal alterations do not emerge as an isolated deficit, but rather as components of dissociable large-scale networks involving medial fronto-parietal\u0026ndash;cerebellar regions (tIV18) and a coupled affective GM\u0026ndash;WM system (tIV05). From a translational perspective, this network-level organization suggests that emotion dysregulation in adolescent BPD reflects an imbalance between affective drive and regulatory control, rather than a uniform impairment of executive functioning. The affective GM\u0026ndash;WM network (tIV05), which is strongly associated with anxiety, self-injurious behaviors, dissociative symptoms, and functional outcome, may represent a primary target for interventions aimed at attenuating maladaptive emotional reactivity, for instance through neuromodulatory approaches targeting prefrontal nodes connected to limbic circuits. Conversely, the medial fronto-parietal\u0026ndash;cerebellar network (tIV18), associated with emotional nonacceptance and reduced cognitive efficiency, highlights a complementary target for interventions aimed at strengthening executive and attentional control processes. Importantly, conceptualizing these alterations at the network level suggests that effective early interventions in adolescent BPD may need to be functionally differentiated and network-specific, targeting affective hyper-responsivity and regulatory control through distinct but complementary mechanisms. This perspective is particularly relevant in adolescence, a developmental period marked by heightened neuroplasticity and ongoing maturation of fronto-parietal control systems.\u003c/p\u003e \u003c/div\u003e "},{"header":"Limitations and conclusions","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003cp\u003eThis study provides robust evidence that adolescent borderline personality disorder is characterized by distinct large-scale structural brain networks, rather than isolated regional abnormalities. By adopting a multimodal, data-driven framework, our findings point to fronto\u0026ndash;cerebellar loop dysfunction as a previously underexplored neurodevelopmental mechanism contributing to emotion dysregulation and self-injurious vulnerability in adolescent borderline personality disorder.\u003c/p\u003e \u003cp\u003eSeveral considerations should be taken into account when interpreting these findings. Participants were recruited from a tertiary psychiatric center and predominantly represented adolescents with moderate-to-severe BPD, characterized by high levels of emotion dysregulation, frequent non-suicidal self-injury, comorbid anxiety, and functional impairment. While this may limit generalizability to milder or subthreshold presentations, it also constitutes a strength of the study, as it ensures high diagnostic reliability and clinical relevance. Exclusion of major psychiatric comorbidities known to exert strong, independent effects on brain structure enhanced internal validity and allowed for a more precise characterization of BPD-related neural alterations. Although comorbid anxiety remained prevalent, this reflects the typical clinical profile of adolescent BPD and underscores the ecological validity of the sample. Finally, while the present study focused on structural MRI, future research should extend this multimodal framework to include functional and longitudinal measures to better characterize developmental trajectories and treatment-related changes.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eConflict of interest\u003c/h2\u003e\n\u003cp\u003eThe authors declare that they have no conflicts of interest\u003c/p\u003e\n\u003ch2\u003eEthical approval\u003c/h2\u003e\n\u003cp\u003eAll procedure complied with the ethical standard of the 1964 Helsinki Declaration Institutional Review Board approval was obtained from our Research Ethics Committee (IRB: 2022020227) Written informed consent was obtained from legal guardians of all adolescent participants\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eAG analyzed the data, visualized the results, and wrote the manuscriptXY, QX conceived this project, wrote and revised the manuscriptXL, XW, JZ, FW, LS, LW wrote and revised the manuscript\u003c/p\u003e\n\u003ch2\u003eAcknowledgment\u003c/h2\u003e\n\u003cp\u003eThe authors have no acknowledgements to declare\u003c/p\u003e\n\u003ch2\u003eData Availability\u003c/h2\u003e\n\u003cp\u003eThe data that support the findings of this study are available on request from the corresponding authors. The data are not publicly available due to their containing information that could compromise the privacy of research participants\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAas, I. H. M. (2011). Guidelines for rating Global Assessment of Functioning (GAF). \u003cem\u003eAnnals of General Psychiatry, 10\u003c/em\u003e(1), 2. https://doi.org/10.1186/1744-859X-10-2 \u003c/li\u003e\n\u003cli\u003eAdalı, T., \u0026amp; Calhoun, V. D. (2015). Multimodal data fusion: An overview of methods, challenges, and prospects. Proceedings of the IEEE, 103(9), 1449\u0026ndash;1477.\u003c/li\u003e\n\u003cli\u003eAlexander-Bloch, A., Giedd, J. N., \u0026amp; Bullmore, E. (2013). Imaging structural co-variance between human brain regions. \u003cem\u003eNature Reviews Neuroscience, 14\u003c/em\u003e(5), 322\u0026ndash;336. https://doi.org/10.1038/nrn3465\u003c/li\u003e\n\u003cli\u003eAshburner, J., \u0026amp; Friston, K. J. (2000). Voxel-based morphometry\u0026mdash;the methods. \u003cem\u003eNeuroImage, 11\u003c/em\u003e(6), 805\u0026ndash;821.\u003c/li\u003e\n\u003cli\u003eBaggio, T., Grecucci, A., Meconi, F., \u0026amp; Messina, I. (2023). Anxious brains: A combined data fusion machine learning approach to predict trait anxiety from morphometric features. \u003cem\u003eSensors, 23\u003c/em\u003e(2).\u003c/li\u003e\n\u003cli\u003eBassett, D. S., Bullmore, E., Verchinski, B. A., Mattay, V. S., Weinberger, D. R., \u0026amp; Meyer-Lindenberg, A. (2008). Hierarchical organization of human cortical networks in health and schizophrenia. \u003cem\u003eJournal of Neuroscience, 28\u003c/em\u003e(37), 9239\u0026ndash;9248. https://doi.org/10.1523/JNEUROSCI.1929-08.2008\u003c/li\u003e\n\u003cli\u003eBirmaher, B., Brent, D. A., Chiappetta, L., Bridge, J., Monga, S., \u0026amp; Baugher, M. (1999). Psychometric properties of the Screen for Child Anxiety Related Emotional Disorders (SCARED): A replication study. \u003cem\u003eJournal of the American Academy of Child \u0026amp; Adolescent Psychiatry, 38\u003c/em\u003e(10), 1230\u0026ndash;1236.\u003c/li\u003e\n\u003cli\u003eBiswal, B. B., Yetkin, F. Z., Haughton, V. M., \u0026amp; Hyde, J. S. (1995). \u003cem\u003eFunctional connectivity in the motor cortex of resting human brain using echo-planar MRI.\u003c/em\u003e Magnetic Resonance in Medicine, 34(4), 537\u0026ndash;541.\u003c/li\u003e\n\u003cli\u003eBohus, M., et al. (2021). Borderline personality disorder. \u003cem\u003eThe Lancet, 398\u003c/em\u003e(10310), 1528\u0026ndash;1540.\u003c/li\u003e\n\u003cli\u003eBozzatello, P., Morese, R., Valentini, M. C., Rocca, P., \u0026amp; Bellino, S. (2021). How interpersonal psychotherapy changes the brain: A study of fMRI in borderline personality disorder. \u003cem\u003eJournal of Clinical Psychiatry, 83\u003c/em\u003e(1). https://doi.org/10.4088/JCP.21m13918\u003c/li\u003e\n\u003cli\u003eBrunner, R., et al. (2010). Reduced prefrontal and orbitofrontal gray matter in female adolescents with borderline personality disorder: Is it disorder-specific? \u003cem\u003eNeuroImage, 49\u003c/em\u003e(1), 114\u0026ndash;120.\u003c/li\u003e\n\u003cli\u003eCalhoun, V. D., Adalı, T., Pearlson, G. D., \u0026amp; Pekar, J. J. (2006). A method for making group inferences from functional MRI data using independent component analysis. Human Brain Mapping, 14(3), 140\u0026ndash;151.\u003c/li\u003e\n\u003cli\u003eCanessa, N., Crespi, C., Motterlini, M., Baud-Bovy, G., Chierchia, G., Pantaleo, G., et al. (2013). The functional and structural neural basis of individual differences in social behavior. \u003cem\u003eJournal of Neuroscience, 33\u003c/em\u003e(36), 14307\u0026ndash;14317.\u003c/li\u003e\n\u003cli\u003eChanen, A. M., et al. (2008). Orbitofrontal, amygdala, and hippocampal volumes in teenagers with first-presentation borderline personality disorder. \u003cem\u003ePsychiatry Research: Neuroimaging, 163\u003c/em\u003e(2), 116\u0026ndash;125.\u003c/li\u003e\n\u003cli\u003eDadomo, H., Grecucci, A., Giardini, I., Ugolini, E., Carmelita, A., \u0026amp; Panzeri, M. (2016). Schema therapy for emotional dysregulation: Theoretical implications and clinical application. \u003cem\u003eFrontiers in Psychology, 7\u003c/em\u003e, 1987. https://doi.org/10.3389/fpsyg.2016.01987\u003c/li\u003e\n\u003cli\u003eDadomo, H., et al. (2018). Neurobiological underpinnings of borderline personality disorder. \u003cem\u003ePsychiatric Research.\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003eDadomo, H., Panzeri, M., Caponcello, D., Carmelita, A., \u0026amp; Grecucci, A. (2018). Schema therapy for emotional dysregulation in personality disorders: A review. \u003cem\u003eCurrent Opinion in Psychiatry, 31\u003c/em\u003e(1), 43\u0026ndash;49.\u003c/li\u003e\n\u003cli\u003eDadomo, H., Salvato, G., Lapomarda, G., Ciftci, Z., Messina, I., \u0026amp; Grecucci, A. (2022). Structural features predict sexual trauma and interpersonal problems in borderline personality disorder but not in controls: A multi-voxel pattern analysis. \u003cem\u003eFrontiers in Human Neuroscience, 23\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eDepping, M. S., Wolf, N. D., Vasic, N., Sambataro, F., Thomann, P. A., \u0026amp; Wolf, R. C. (2016). Common and distinct structural network abnormalities in major depressive disorder and bipolar disorder. \u003cem\u003eProgress in Neuropsychopharmacology and Biological Psychiatry, 65\u003c/em\u003e, 127\u0026ndash;133.\u003c/li\u003e\n\u003cli\u003eDing, J. B., \u0026amp; Hu, K. (2021). Structural MRI brain alterations in borderline personality disorder and bipolar disorder. \u003cem\u003eCureus, 13\u003c/em\u003e, e16425. https://doi.org/10.7759/cureus.16425\u003c/li\u003e\n\u003cli\u003eEllison, W. D., et al. (2018). Community and clinical epidemiology of borderline personality disorder. \u003cem\u003ePsychiatric Clinics of North America, 41\u003c/em\u003e(4), 561\u0026ndash;573.\u003c/li\u003e\n\u003cli\u003eEvans, A. C. (2013). Networks of anatomical covariance. \u003cem\u003eNeuroImage, 80\u003c/em\u003e, 489\u0026ndash;504. https://doi.org/10.1016/j.neuroimage.2013.05.054\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eFettes, P., Schulze, L., \u0026amp; Downar, J.\u003c/strong\u003e (2017). Cortico-striatal-thalamic loop circuits of the orbitofrontal cortex: Promising therapeutic targets in psychiatric illness. \u003cem\u003eFrontiers in Systems Neuroscience, 11\u003c/em\u003e, 25. https://doi.org/10.3389/fnsys.2017.00025\u003c/li\u003e\n\u003cli\u003eFox, M. D., Snyder, A. Z., Vincent, J. L., Corbetta, M., Van Essen, D. C., \u0026amp; Raichle, M. E. (2005). \u003cem\u003eThe human brain is intrinsically organized into dynamic, anticorrelated functional networks.\u003c/em\u003e Proceedings of the National Academy of Sciences of the USA, 102(27), 9673\u0026ndash;9678. doi:10.1073/pnas.0504136102 \u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eFrederickson, J. J., Messina, I., \u0026amp; Grecucci, A.\u003c/strong\u003e (2018). Dysregulated anxiety and dysregulating defenses: Toward an emotion regulation informed dynamic psychotherapy. \u003cem\u003eFrontiers in Psychology, 9\u003c/em\u003e, 2054. https://doi.org/10.3389/fpsyg.2018.02054\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGhomroudi, P. A., Scaltritti, M., \u0026amp; Grecucci, A.\u003c/strong\u003e (2023). Decoding reappraisal and suppression from neural circuits: A combined supervised and unsupervised machine learning approach. \u003cem\u003eCognitive, Affective, \u0026amp; Behavioral Neuroscience, 23\u003c/em\u003e, 1095\u0026ndash;1112. https://doi.org/[Insert DOI]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGiuliani, N. R., Drabant, E. M., \u0026amp; Gross, J. J.\u003c/strong\u003e (2011). Anterior cingulate cortex volume and emotion regulation: Is bigger better? \u003cem\u003eBiological Psychology, 86\u003c/em\u003e(3), 379\u0026ndash;382. https://doi.org/10.1016/j.biopsycho.2010.11.010\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGoethals, I., Audenaert, K., Jacobs, F., Van den Eynde, F., Bernagie, K., Kolindou, A., ... \u0026amp; Van Heeringen, C.\u003c/strong\u003e (2005). Brain perfusion SPECT in impulsivity-related personality disorders. \u003cem\u003eBehavioural Brain Research, 157\u003c/em\u003e(1), 187-192. https://doi.org/10.1016/j.bbr.2004.06.022\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGoodman, M., Hazlett, E. A., Avedon, J. B., Siever, D. R., Chu, K. W., \u0026amp; New, A. S.\u003c/strong\u003e (2011). Anterior cingulate volume reduction in adolescents with borderline personality disorder and co-morbid major depression. \u003cem\u003eJournal of Psychiatric Research, 45\u003c/em\u003e(6), 803-807. https://doi.org/10.1016/j.jpsychires.2010.11.011\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGrecucci, A., Dadomo, H., Salvato, G., Lapomarda, G., Sorella, S., \u0026amp; Messina, I.\u003c/strong\u003e (2023). Abnormal brain circuits characterize borderline personality and mediate the relationship between specific childhood traumas and symptoms: An mCCA+ jICA and random forest approach. \u003cem\u003eSensors, 23\u003c/em\u003e(5).\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGrecucci, A., Lapomarda, G., Messina, I., Monachesi, B., Sorella, S., \u0026amp; Siugzdaite, R.\u003c/strong\u003e (2022). Structural features related to affective instability correctly classify the diagnosis of borderline personality disorder: A supervised machine learning approach. \u003cem\u003eFrontiers in Psychiatry\u003c/em\u003e. https://doi.org/10.3389/fpsyt.2022.804440\u003c/li\u003e\n\u003cli\u003eGroves, A. R., Beckmann, C. F., Smith, S. M., \u0026amp; Woolrich, M. W. (2011). Linked independent component analysis for multimodal data fusion. NeuroImage, 54(3), 2198\u0026ndash;2217. doi:10.1016/j.neuroimage.2010.09.073\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eHallett, M., et al.\u003c/strong\u003e (2016). Cerebellar contributions to emotional processing. \u003cem\u003eCerebellum\u003c/em\u003e. https://doi.org/[Insert DOI]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eHazlett, E. A., New, A. S., Newmark, R., Haznedar, M. M., Lo, J. N., Speiser, L. J., ... \u0026amp; Buchsbaum, M. S.\u003c/strong\u003e (2005). Reduced anterior and posterior cingulate gray matter in borderline personality disorder. \u003cem\u003eBiological Psychiatry, 58\u003c/em\u003e(8), 614-623. https://doi.org/10.1016/j.biopsych.2005.04.029\u003c/li\u003e\n\u003cli\u003eHebart, M. N., \u0026amp; Baker, C. I. (2018). Deconstructing multivariate decoding for the study of brain function. NeuroImage, 180, 4\u0026ndash;18.\u003c/li\u003e\n\u003cli\u003eHyv\u0026auml;rinen, A., \u0026amp; Oja, E. (2000). Independent component analysis: Algorithms and applications. \u003cem\u003eNeural Networks, 13\u003c/em\u003e(4\u0026ndash;5), 411\u0026ndash;430.\u003c/li\u003e\n\u003cli\u003eIrle, E., et al. (2007). Size abnormalities of the superior parietal cortices are related to dissociation in borderline personality disorder. \u003cem\u003ePsychiatry Research, 156\u003c/em\u003e(2), 139\u0026ndash;149.\u003c/li\u003e\n\u003cli\u003eJin, X., et al. (2016). A voxel-based morphometric MRI study in young adults with borderline personality disorder. \u003cem\u003ePLOS ONE, 11\u003c/em\u003e(1), e0147938.\u003c/li\u003e\n\u003cli\u003eKherif, F., \u0026amp; Latypova, A. (2020). Principal component analysis. In Mechelli \u0026amp; Vieira (Eds.), Machine Learning: Methods and Applications to Brain Disorders (Academic Press).\u003c/li\u003e\n\u003cli\u003eKoenigsberg, H. W., Anwunah, I., New, A. S., Mitropoulou, V., Schopick, F., \u0026amp; Siever, L. J. (1999). Relationship between depression and borderline personality disorder. \u003cem\u003eDepression and Anxiety, 10\u003c/em\u003e, 158\u0026ndash;167. https://doi.org/10.1002/(SICI)1520-6394(1999)10:4\u0026lt;158::AID-DA4\u0026gt;3.0.CO;2-B\u003c/li\u003e\n\u003cli\u003eKoziol, L. F., et al. (2014). The cerebellum and emotion: A review of recent findings. \u003cem\u003eNeuropsychology Review\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eLangerbeck, M., Baggio, T., Messina, I., Bhat, S., \u0026amp; Grecucci, A. (2023). Borderline shades: Morphometric features predict borderline personality traits but not histrionic traits. \u003cem\u003eNeuroImage: Clinical, 40\u003c/em\u003e, 103530. https://doi.org/10.1016/j.nicl.2023.103530\u003c/li\u003e\n\u003cli\u003eLapomarda, G., Grecucci, A., Messina, I., Pappaianni, E., \u0026amp; Dadomo, H. (2021). Common and different gray and white matter alterations in bipolar and borderline personality disorder: A source-based morphometry study. \u003cem\u003eBrain Research, 1762\u003c/em\u003e, 147401. https://doi.org/10.1016/j.brainres.2021.147401\u003c/li\u003e\n\u003cli\u003eLapomarda, G., Pappaianni, E., Siugzdaite, R., Sanfey, A. G., Rumiati, R. I., \u0026amp; Grecucci, A. (2021). Out of control: An altered parieto-occipital-cerebellar network for impulsivity in bipolar disorder. \u003cem\u003eBehavioral Brain Research, 406\u003c/em\u003e, 113228. https://doi.org/10.1016/j.bbr.2021.113228\u003c/li\u003e\n\u003cli\u003eLeBoeuf, A., Guil\u0026eacute;, J. M., Labelle, R., \u0026amp; Luck, D. (2016). Functional neuroimaging pilot study of borderline personality disorder in adolescents. \u003cem\u003eSante Mentale au Quebec, 41\u003c/em\u003e(1), 141\u0026ndash;162.\u003c/li\u003e\n\u003cli\u003eLee, J.-H., Lee, T.-W., Jolesz, F. A., \u0026amp; Yoo, S.-S. (2008). \u003cem\u003eIndependent vector analysis (IVA): Multivariate approach for fMRI group study.\u003c/em\u003e NeuroImage, 40(1), 86\u0026ndash;109. doi:10.1016/j.neuroimage.2007.11.019 \u003c/li\u003e\n\u003cli\u003eLevine, A. Z., et al. (2020). Nonsuicidal self-injury and suicide: Differences between those with and without borderline personality disorder. \u003cem\u003eJournal of Personality Disorders, 34\u003c/em\u003e(1), 131\u0026ndash;144.\u003c/li\u003e\n\u003cli\u003eLi, J., et al. (2018). Psychometric properties of the Chinese version of the Difficulties in Emotion Regulation Scale (DERS): Factor structure, reliability, and validity. \u003cem\u003ePsychological Assessment, 30\u003c/em\u003e(5), e1\u0026ndash;e9.\u003c/li\u003e\n\u003cli\u003eLuo, X., Mao, Q., Shi, J., Wang, X., \u0026amp; Li, C. R. (2019). Putamen gray matter volumes in neuropsychiatric and neurodegenerative disorders. \u003cem\u003eWorld Journal of Psychiatry and Mental Health Research, 3\u003c/em\u003e, 1020.\u003c/li\u003e\n\u003cli\u003eMartin, J., et al. (2013). Psychometric properties of the functions and addictive features scales of the Ottawa Self-Injury Inventory: A preliminary investigation using a university sample. \u003cem\u003ePsychological Assessment, 25\u003c/em\u003e(3), 1013\u0026ndash;1018.\u003c/li\u003e\n\u003cli\u003eMcKeown, M. J., \u0026amp; Sejnowski, T. J. (1998). Independent component analysis of fMRI data: Examining the assumptions. \u003cem\u003eHuman Brain Mapping, 6\u003c/em\u003e(5\u0026ndash;6), 368\u0026ndash;372.\u003c/li\u003e\n\u003cli\u003eMirkovic, B., et al. (2021). Borderline personality disorder and adolescent suicide attempt: The mediating role of emotional dysregulation. \u003cem\u003eBMC Psychiatry, 21\u003c/em\u003e(1), 393.\u003c/li\u003e\n\u003cli\u003ePappaianni, E., Siugzdaite, R., Vettori, S., Venuti, P., Job, R., \u0026amp; Grecucci, A. (2018). Three shades of grey: Detecting brain abnormalities in children with autism by using source-, voxel- and surface-based morphometry. \u003cem\u003eEuropean Journal of Neuroscience, 47\u003c/em\u003e(6), 690\u0026ndash;700.\u003c/li\u003e\n\u003cli\u003eParis, J., Gunderson, J., \u0026amp; Weinberg, I. (2007). The interface between borderline personality disorder and bipolar spectrum disorders. \u003cem\u003eComprehensive Psychiatry, 48\u003c/em\u003e, 145\u0026ndash;154. https://doi.org/10.1016/j.comppsych.2006.10.001\u003c/li\u003e\n\u003cli\u003ePearlson, G. D., Calhoun, V. D., \u0026amp; Liu, J. (2015). An introductory review of parallel independent component analysis (p-ICA) and a guide to applying p-ICA to genetic data and imaging phenotypes to identify disease-associated biological pathways and systems in common complex disorders. \u003cem\u003eFrontiers in Genetics, 6\u003c/em\u003e, 276.\u003c/li\u003e\n\u003cli\u003eQuattrini, G., Magni, L. R., Lanfredi, M., Pedrini, L., Carcione, A., Riccardi, I., Corbo, D., Gasparotti, R., Rossi, R., Pievani, M., \u0026amp; Caroli, A. (2022). Aberrant structural connectivity of the triple network system in borderline personality disorder is associated with behavioral dysregulation. \u003cem\u003eJournal of Clinical Medicine, 11\u003c/em\u003e(7). https://doi.org/10.3390/jcm11071757\u003c/li\u003e\n\u003cli\u003eRichter, J., et al. (2014). Reduced cortical and subcortical volumes in female adolescents with borderline personality disorder. \u003cem\u003ePsychiatry Research, 221\u003c/em\u003e(3), 179\u0026ndash;186.\u003c/li\u003e\n\u003cli\u003eRossi, R., Pievani, M., Lorenzi, M., Boccardi, M., Beneduce, R., Bignotti, S., et al. (2013). Structural brain features of borderline personality and bipolar disorders. \u003cem\u003ePsychiatry Research, 213\u003c/em\u003e, 83\u0026ndash;91. https://doi.org/10.1016/j.pscychresns.2012.07.002\u003c/li\u003e\n\u003cli\u003eRuocco, A. C., Amirthavasagam, S., Choi-Kain, L. W., \u0026amp; McMain, S. F. (2013). Neural correlates of negative emotionality in borderline personality disorder: An activation-likelihood-estimation meta-analysis. \u003cem\u003eBiological Psychiatry, 73\u003c/em\u003e, 153\u0026ndash;160. https://doi.org/10.1016/j.biopsych.2012.07.014\u003c/li\u003e\n\u003cli\u003eSalvador, R., et al. (2011). Cerebellar contributions to anxiety and depression. \u003cem\u003eNeuroscience\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eSaviola, F., Pappaianni, E., Monti, A., Grecucci, A., Jovicich, J., \u0026amp; de Pisapia, N. (2020). Trait and state anxiety are mapped differently in the human brain. \u003cem\u003eScientific Reports, 10\u003c/em\u003e(1). https://doi.org/10.1038/s41598-020-68008-z\u003c/li\u003e\n\u003cli\u003eSchnack, H. G. (2020). Improving individual predictions: Machine learning in psychiatry. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 5(6), 560\u0026ndash;571.\u003c/li\u003e\n\u003cli\u003eSchmahmann, J. D. (2010). The role of the cerebellum in emotion and cognition. \u003cem\u003eCerebellum\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eSchulze, L., Schmahl, C., \u0026amp; Niedtfeld, I. (2016). Neural correlates of disturbed emotion processing in borderline personality disorder: A multimodal meta-analysis. \u003cem\u003eBiological Psychiatry, 79\u003c/em\u003e, 97\u0026ndash;106. https://doi.org/10.1016/j.biopsych.2015.03.027\u003c/li\u003e\n\u003cli\u003eSchulze, L., et al. (2016). The cerebellum and emotion dysregulation in borderline personality disorder. \u003cem\u003eFrontiers in Psychology\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eSeeley, W. W., Crawford, R. K., Zhou, J., Miller, B. L., \u0026amp; Greicius, M. D. (2009). Neurodegenerative diseases target large-scale human brain networks. \u003cem\u003eNeuron, 62\u003c/em\u003e(1), 42\u0026ndash;52. https://doi.org/10.1016/j.neuron.2009.03.024\u003c/li\u003e\n\u003cli\u003eSoloff, P., Nutche, J., Goradia, D., \u0026amp; Diwadkar, V. (2008). Structural brain abnormalities in borderline personality disorder: A voxel-based morphometry study. \u003cem\u003ePsychiatry Research, 164\u003c/em\u003e, 223\u0026ndash;236. https://doi.org/10.1016/j.pscychresns.2008.02.003\u003c/li\u003e\n\u003cli\u003eSoloff, P. H., Kelly, T. M., Strotmeyer, S. J., Malone, K. M., \u0026amp; Mann, J. J. (2003). Impulsivity, gender, and response to fenfluramine challenge in borderline personality disorder. \u003cem\u003ePsychiatry Research, 119\u003c/em\u003e(1-2), 11\u0026ndash;24. https://doi.org/10.1016/s0165-1781(03)00100-8\u003c/li\u003e\n\u003cli\u003eSorella, S., Lapomarda, G., Messina, I., Frederickson, J. J., Siugzdaite, R., Job, R., et al. (2019). Testing the expanded continuum hypothesis of schizophrenia and bipolar disorder. Neural and psychological evidence for shared and distinct mechanisms. \u003cem\u003eNeuroImage, 23\u003c/em\u003e, 101854. https://doi.org/10.1016/j.nicl.2019.101854\u003c/li\u003e\n\u003cli\u003eSui, J., Adalı, T., Yu, Q., Chen, J., \u0026amp; Calhoun, V. D. (2012). A review of multivariate methods for multimodal fusion of brain imaging data. Journal of Neuroscience Methods, 204(1), 68\u0026ndash;81. doi:10.1016/j.jneumeth.2011.10.031\u003c/li\u003e\n\u003cli\u003eVieira, S., Pinaya, W. H. L., \u0026amp; Mechelli, A. (2020). Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders. Neuroscience \u0026amp; Biobehavioral Reviews, 113, 546\u0026ndash;560.\u003c/li\u003e\n\u003cli\u003eWechsler, D. (1999). \u003cem\u003eWechsler abbreviated scale of intelligence\u003c/em\u003e. The Psychological Corporation.\u003c/li\u003e\n\u003cli\u003eWhittle, S., Chanen, A. M., Fornito, A., McGorry, P. D., Pantelis, C., \u0026amp; Y\u0026uuml;cel, M. (2009). Anterior cingulate volume in adolescents with first-presentation borderline personality disorder. \u003cem\u003ePsychiatry Research, 172\u003c/em\u003e(2), 155\u0026ndash;160. https://doi.org/10.1016/j.pscychresns.2008.12.004\u003c/li\u003e\n\u003cli\u003eWood, A., Kroll, L., Moore, A., \u0026amp; Harrington, R. (1995). Properties of the Mood and Feelings Questionnaire in adolescent psychiatric outpatients: A research note. \u003cem\u003eJournal of Child Psychology and Psychiatry, 36\u003c/em\u003e(2), 327\u0026ndash;334.\u003c/li\u003e\n\u003cli\u003eXiao, Q., Shen, L., He, H., et al. (2024). Alteration of prefrontal cortex and its associations with emotional and cognitive dysfunctions in adolescent borderline personality disorder. \u003cem\u003eEuropean Child and Adolescent Psychiatry, 33\u003c/em\u003e, 3937\u0026ndash;3949. https://doi.org/10.1007/s00787-024-02438-2\u003c/li\u003e\n\u003cli\u003eXu, L., Groth, K. M., Pearlson, G., Schretlen, D. J., \u0026amp; Calhoun, V. D. (2009). Source-based morphometry: The use of independent component analysis to identify gray matter differences with application to schizophrenia. \u003cem\u003eHuman Brain Mapping, 30\u003c/em\u003e(3), 711\u0026ndash;724.\u003c/li\u003e\n\u003cli\u003eXu, T., Cullen, K. R., Mueller, B., Schreiner, M. W., Lim, K. O., Schulz, S. C., et al. (2016). Network analysis of functional brain connectivity in borderline personality disorder using resting-state fMRI. \u003cem\u003eNeuroImage, 11\u003c/em\u003e, 302\u0026ndash;315. https://doi.org/10.1016/j.nicl.2016.02.006\u003c/li\u003e\n\u003cli\u003eYi, X., Fu, Y., Ding, J., Jiang, F., Han, Z., Zhang, Y., et al. (2024). Altered gray matter volume and functional connectivity in adolescent borderline personality disorder with non-suicidal self-injury behavior. \u003cem\u003eEuropean Child and Adolescent Psychiatry, 33\u003c/em\u003e(1), 193\u0026ndash;202. https://doi.org/10.1007/s00787-023-02161-4\u003c/li\u003e\n\u003cli\u003eYi, X., Fu, Y., Zhang, Z., Jiang, F., Xiao, Q., \u0026amp; Chen, B. T. (2023). Altered regional homogeneity and its association with cognitive function in adolescents with borderline personality disorder. \u003cem\u003eJournal of Psychiatry and Neuroscience, 48\u003c/em\u003e(1), E1\u0026ndash;E10. https://doi.org/10.1503/jpn.220144\u003c/li\u003e\n\u003cli\u003eYi, X., Wang, X., Shen, L., Fu, Y., Zhang, Z., Ding, J., Jiang, F., Zhang, Z., Grecucci, A., \u0026amp; Chen, B. T. (2025). \u003cem\u003eWhite matter microstructural abnormalities related to emotional dysfunction and childhood trauma characterize adolescents with borderline personality disorder\u003c/em\u003e. Journal of Affective Disorders. Advance online publication. https://doi.org/10.1007/s11682-025-01046-1 \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8860480/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8860480/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eIntroduction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrevious work demonstrated that adolescents with borderline personality disorder (aBPD) exhibit gray and white matter density (GM) alterations in multiple regions, yet prior neuroimaging studies have predominantly examined GM and WM in isolation, limiting the ability to capture their joint neurodevelopmental alterations. The present study aims to overcome this limitation by integrating GM and WM structural features within a unified unsupervised multimodal data-fusion framework, in order to identify latent brain components reflecting shared and dissociable patterns and to relate these multimodal signatures to clinical and functional measures. We predicted that adolescents with BPD would show coupled and decoupled gray-white matter networks related to affective processing and executive functions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHigh-resolution T1-weighted structural MRI data from 129 adolescents with borderline personality disorder (aged 12–17) and 107 age-, gender-, and education-matched healthy controls were analyzed using transposed independent vector analysis (tIVA-G), integrating gray and white matter features. Multiple clinical measures were collected and used to assess the clinical relevance of the identified multimodal components. In addition, decision tree models were applied to derive an interpretable predictive model discriminating adolescents with BPD from healthy controls.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo distinct structural components differentiating adolescents with BPD from healthy controls were identified. A first component (IC18) reflected a gray matter-driven pattern, characterized by reduced volume in a fronto-cerebellar loop, with no corresponding white matter contribution. A second component (IC5) captured a coupled gray matter–white matter alteration, showing increased gray matter in limbic–affective regions and Default network regions, including temporal and medial frontal areas, precuneus, and cerebellum, alongside reduced white matter in frontal regions. This network was significantly associated with greater emotion dysregulation, self-injurious behaviors, internalizing symptoms, and lower global functioning. Decision tree discriminated adolescents with BPD from healthy controls with moderate accuracy (71.7%).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTogether, these multimodal patterns link large-scale brain organization to emotion dysregulation, self-injurious behaviors, and functional impairment in youth with BPD, providing a more comprehensive neurobiological account of the disorder and informing future biomarker-oriented and mechanistically grounded interventions.\u003c/p\u003e","manuscriptTitle":"Coupled and decoupled fronto–cerebellar and affective brain networks characterize adolescents with borderline personality disorder","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-08 17:16:21","doi":"10.21203/rs.3.rs-8860480/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"242115e3-0bca-461d-91d7-cf0b47f10b1e","owner":[],"postedDate":"March 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-13T08:24:12+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-08 17:16:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8860480","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8860480","identity":"rs-8860480","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}