EEG Microstates and Functional Connectivity Abnormalities in Depressive Disorders: Diagnostic Potential and Machine Learning Application

EEG Microstates and Functional Connectivity Abnormalities in Depressive Disorders: Diagnostic Potential and Machine Learning Application | 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 EEG Microstates and Functional Connectivity Abnormalities in Depressive Disorders: Diagnostic Potential and Machine Learning Application Cun Li, Ke Shi, Yanhui Song, Ye Xia, Ziwei Wang, Jie Feng, Kai Wang, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6170701/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 13 You are reading this latest preprint version Abstract Background: Diagnosis of depressive disorders (DD) primarily heavily relies on clinical interviews and subjective symptom reporting, highlighting the urgent need for objective biomarkers. This study aimed to identify novel electroencephalogram (EEG) features, utilizing high-density EEG and source localization techniques, to aid in the diagnosis of DD. Methods: Resting-state EEG data were collected from 115 patients with DD and 43 healthy controls (HCs). Microstate analysis and functional connectivity (FC) analysis were performed to extract EEG features. Statistical analyses were conducted to compare group differences, and a support vector machine (SVM) algorithm was applied to evaluate the diagnostic accuracy of these features. Results: Compared to HCs, DD patients showed a significant increased transition probability from microstate D to B (PFDR = 0.048). Additionally, significant elevations in FC within specific regions of the default mode network (DMN) were observed in the delta-band, theta-band, and beta-band (P < 0.05), as well as between parts of the DMN and the salience network (SN) in the theta-band and alpha-band (P 0.05). The classification accuracies for distinguishing DD patients from HCs using the SVM classifier were 66.7%, 76.2%, and 81.0% based on microstate features, FC features, and a combination of both, respectively. Conclusions: Patients with DD exhibited distinctive microstates and atypical alterations in brain network connectivity. Integrating these features with machine learning algorithms offers a promising approach to improving the objective diagnosis of DD. Trial registration: The study was registered on http://www.chictr.org.cn/ and the registration number was ChiCTR2200057365 (registration date: March 9, 2022). Depressive disorders Electroencephalogram Microstates Functional connectivity Support vector machine Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Depression disorder (DD) is a prevalent and debilitating mental health condition characterized by persistent low mood, anhedonia, and a range of cognitive, emotional, and somatic symptoms[ 1 ]. According to the World Health Organization (WHO), DD affects approximately 3.8% of the global population, with a particularly high prevalence of 5.0% among adults[ 1 – 3 ]. Despite its prevalence and impact, diagnosing DD is challenging due to the heavy reliance on clinical interviews and subjective symptom reporting, lacking objective biomarkers[ 4 , 5 ]. This highlights the urgent need to identify reliable and objective markers to enhance diagnostic accuracy and guide personalized treatment strategies. Prior research has indicated that aberrant brain structure and function are underlying the pathophysiological mechanisms of DD[ 6 ]. A variety of neuroimaging techniques, including electroencephalogram (EEG) [ 7 , 8 ], functional magnetic resonance imaging (fMRI)[ 9 , 10 ], positron emission tomography (PET)[ 11 ], and magnetoencephalography (MEG)[ 12 ], have been widely employed to investigate these mechanisms. Among these methods, EEG has gained considerable attention due to its ability to directly measure neuronal activity, coupled with its high temporal resolution and portability[ 4 ]. Moreover, advanced computational methods, such as microstate analysis and functional connectivity (FC) analysis, have been developed to extract meaningful features from EEG signals, providing novel insights into the dynamic spatiotemporal patterns of brain activity [ 13 , 14 ]. Microstates, which are brief, collections of EEG topographic maps that typically dominate for 80-120ms before transitioning to another map, offer valuable insights into the dynamics of large-scale brain networks[ 13 ]. Research has consistently identified four canonical microstates (A, B, C, and D), which are closely related to specific resting-state brain networks[ 15 , 16 ]. Studies have reported abnormal microstate patterns in DD, such as altered durations, occurrences, and coverages of specific microstates[ 17 , 18 ]. Moreover, microstate features have demonstrated potential as diagnostic biomarkers for DD when combined with machine learning algorithms[ 19 ]. FC analysis is another widely used approach to investigate the interactions among brain regions and networks[ 20 ]. While fMRI-based FC studies have provided valuable insights into the network abnormalities in DD, they are limited by the low temporal resolution of fMRI, which typically measures brain activity on a timescale of seconds[ 21 ]. In contrast, EEG allowing for the investigation of FC at shorter timescales[ 4 ], which is particularly important for understanding the rapid fluctuations in brain network interactions that underlie cognitive and affective processes often impaired in DD[ 4 ]. Machine learning and feature selection methods have been utilized to differentiate DD patients from HCs based on FC[ 22 , 23 ]. For example, Chen et al. employed dynamic EEG FC features to classify DD patients with and without psychotic symptoms, schizophrenia patients, and HCs, obtaining an accuracy of 73.1% in the four-group classification[ 23 ]. While both microstate and FC features have shown potential in distinguishing DD patients from HCs, prior research has been limited by inconsistent findings, small sample sizes and a lack of comprehensive analysis that combining both microstate and FC for improved classification. To address these limitations, the present study set out to test the following hypotheses: (1) DD patients exhibit distinctive EEG microstate features and widespread abnormal brain network connectivity; and (2) The combination of microstate and FC features, analyzed with machine learning algorithms, can more effectively differentiate DD patients from HCs. By utilizing high-density EEG and source localization techniques, this study seeks to enhance the spatial and temporal resolution of EEG-based biomarkers, thereby providing a more comprehensive understanding of DD's pathophysiological mechanisms. Our findings may not only deepen our understanding of the neural mechanisms underlying DD but also facilitate the development of more targeted and effective interventions for this prevalent and debilitating disorder. Materials and Methods Participants This present study enrolled 115 untreated patients diagnosed with DD from Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology from March 2022 to September 2023. The eligibility criteria for participants were as follows: (1) age range between 18 and 65 years; (2) aligning with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) criteria for diagnosing DD; (3) a Hamilton Depression Rating Scale-21 (HAMD-21) score of 17 or higher; (4) a minimum education level equivalent to elementary school; (5) right-handedness; and (6) native Chinese speaker. The exclusion criteria encompassed: (1) a history of mental illnesses such as schizophrenia, severe DD with suicidal ideation, alcohol dependence, or substance misuse; (2) the presence of significant or unstable medical conditions; (3) any current or past head trauma, central nervous system disorders, or other conditions listed under the International Classification of Diseases, Tenth Revision (ICD-10); (4) contraindications to antidepressant medications; (5) use (within the past month) of antidepressants or long-acting antipsychotic medications; (6) impairments such as aphasia, hearing loss, visual impairment, or cognitive dysfunction; and (7) pregnancy, breastfeeding, or plans to conceive during the study period. Additionally, 43 HCs matched for age, gender, and education level were included. These participants met the same exclusion criteria, with the added requirement of no history of psychiatric disorders. A flowchart illustrating the recruitment process is presented in Fig. 1 . Informed consent was obtained following the principles outlined in the Declaration of Helsinki. The study protocol received ethical approval from the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (approval number: TJ-IRB20220205). The study was registered on http://www.chictr.org.cn/ and the registration number was ChiCTR2200057365 (registration date: March 9, 2022). Data Collection Firstly, general demographic data were collected, including age, gender, educational level, marital status, and Body Mass Index (BMI). Educational level was categorized into three tiers: primary school or below, secondary school/vocational school, and college/university degree or higher. Marital status was classified as married or single/divorced/widowed. In addition, psychological assessment data were collected using the HAMD-21, Hamilton Anxiety Rating Scale (HAMA), Childhood Trauma Questionnaire (CTQ), Social Support Rate Scale (SSRS), Patient Health Questionnaire-15 (PHQ-15), Pittsburgh Sleep Quality Index (PSQI), Chinese Perceived Stress Scale (CPSS), Temporal Experience of Pleasure Scale (TEPS), and Digit Symbol Substitution Test (DSST). EEG Recording and Preprocessing Rest-state EEG signals were recorded during 12-minute sessions that alternated between eyes-open and eyes-closed conditions. Recordings took place in a dimly lit, electrically shielded, and soundproof chamber using a 128-channel EEG system (BrainVision Recorder software, Brain Products GmbH, Germany). During the eyes-open condition, participants were instructed to stay as still as possible, avoid blinking or making eye movements, and focus on a central fixation cross. The electrodes were arranged following the international 10/5 system, with a sampling rate of 1000 Hz and the FCz electrode serving as the reference. Impedance levels were maintained below 20 KΩ throughout the recording process. EEG data preprocessing was performed using BrainVision Analyzer software (version 2.2, Brain Products GmbH, Germany) and included the following steps: (1) Re-referencing: Raw EEG data were re-referenced offline to the average of the mastoid electrodes (TP9 and TP10). (2) Filtering: Data were filtered using infinite impulse response (IIR) filters (0.5–45 Hz) and a 50 Hz notch filter. (3) Down sampling: The sampling frequency of rest-state EEG data was reduced to 500Hz. (4) Ocular correction: Eyeblink and ocular movement artifacts were corrected using independent component analysis (ICA) based on established standards[ 24 ]. (5) Artifact rejection: Segments with voltage changes exceeding 50µV between sample points, a voltage difference of 300µV within a segment, or a maximum voltage difference of less than 0.5µV within 100ms were automatically rejected, with interpolation for electrodes with more than 20% bad segments. (6) Segmentation: The cleaned resting-state EEG data was segmented into 2-second trails. Microstates Analysis Microstate analysis was conducted using the Microstates toolbox[ 25 ] integrated within the EEGLAB plugin[ 26 ] in MATLAB R2023a. First, the global field power (GFP) was calculated for each participant at every time point. Given that the local maxima of the GFP curve have the strongest signal strength and highest signal-to-noise ratio, the corresponding topographical maps were selected for subsequent clustering analysis. Next, a modified k-means algorithm was utilized to cluster the topographical maps, yielding four distinct microstates (A-D) for each group. The group-level microstates were then used as template maps. Each participant's topographical maps were assigned to one of the four microstate classes (A-D) based on spatial similarity with the templates. Finally, key microstate parameters, including mean duration, occurrence, coverage, transition probability, and topographic shape, were extracted for each participant. Source Localization Analyses FC derived from scalp electrodes is susceptible to volume conduction effects, which can produce artificial connections[ 27 ]. To address this limitation, calculating FC in the source space can greatly improve accuracy[ 27 ]. In this study, we performed source localization analysis using low-resolution electromagnetic tomography analysis (LORETA) within the BrainVision Analyzer software (version 2.2, Brain Products GmbH, Germany) in the present study[ 28 , 29 ]. LORETA estimates the current density (µA/mm 2 ) of intracranial sources that generate scalp-recorded EEG signals[ 30 ]. Regions of interest (ROIs) were defined based on a publicly available source file ( www.brainproducts.com/files/public/downloads/LORETA/All_ROIs.xml ). The source space encompasses 88 regions, including the bilateral amygdala, bilateral Brodmann areas (BA) (BA1-11, BA13, BA17-25, BA27-47), and bilateral hippocampus[ 31 ], as illustrated in Supplemental Figure S1 . Functional Connectivity In this study, the Phase Lag Index (PLI) was chosen to assess the strength of FC between different ROIs[ 32 ]. PLI, a phase-based connectivity analysis method, is predicted on the distribution of phase angle differences between electrodes pairs[ 32 ]. The underlying principle is that when neural clusters are functionally coupled, their oscillatory sequences become phase-synchronized. For a more comprehensive understanding of brain connectivity, EEG signals were segmented into four frequency sub-bands: delta (δ, 1–4 Hz), theta (θ, 4–8 Hz), alpha (α, 8–12 Hz), and beta (β, 13–30 Hz). The FC values between each pair of ROIs were subsequently calculated across these frequency bands. Statistical Analysis The general characteristics of the study population were summarized as follows: as categorical variables were presented as percentages, continuous variables with normal distributions were reported as means ± standard deviations (SDs), and continuous variables with non-normal distributions were described using medians (interquartile ranges, IQRs). Comparisons between the DD and HC groups were performed using chi-square tests or Fisher's exact tests for categorical variables. For continuous variables, independent two-sample t-tests were employed for normally distributed data, and the Mann-Whitney U test for non-normally distributed data. All statistical analyses were conducted using SPSS (IBM, version 26.0), with the false discovery rate (FDR) correction applied to control for multiple comparisons. Statistical significance was determined using two-tailed tests with a significance level of α = 0.05. Additionally, topographic analysis of variance (TANOVA) based on Rugu software was implemented to explore differences in microstate topographies between the DD and HC groups. TANOVA is a robust, hypothesis-free randomization test that facilitates the statistical comparison of EEG scalp field topographies across two or more conditions[ 33 ]. Furthermore, support vector machine (SVM) algorithms were employed to assess the accuracy of microstate and FC features in distinguishing DD patients from HCs. The steps were as follows: (1) Feature matrix construction: For each frequency band, the FC matrix (88×88) was symmetric; hence, the diagonal elements were removed, and the upper triangular elements of the matrix were extracted as classification features, resulting in 88 × (88 − 1)/2 dimensions for each frequency band. (2) Feature selection: F-score values were utilized to evaluate the correlation of each feature with the disease classification, allowing for dimensionality reduction by retaining the most discriminative features while discarding less informative ones[ 34 ]. (3) Feature set construction: Feature set 1 comprised microstate features (33 dimensions); feature set 2 included FC features (15,312 dimensions); and feature set 3 included both microstate and FC features (15,345 dimensions). (4) Training and testing set division: The stratified holdout method was applied, dividing the data into a training set (80%) and a testing set (20%); 5) SVM model: The SVM algorithm was implemented to address the binary classification problem; 6) Model evaluation: The classifier's generalization ability was quantified using metrics such as accuracy, sensitivity, specificity, receiver operating characteristic (ROC) curves, and the area under the curve (AUC). Results Study participants The study enrolled 115 DD patients and 43 HCs from March 2022 to September 2023. After excluding DD patients with missing baseline EEG data (n = 6), a history of antidepressant use (n = 23), and poor-quality EEG data (n = 22), along with two HCs due two poor-quality EEG data, 69 DD patients and 41 HCs were included in the final analysis. The participant selection process is shown in Figure 1. Baseline characteristics As presented in Table 1, the DD group had a significantly lower BMI (21.83 ± 2.94 kg/m² vs. 23.66 ± 2.63 kg/m², P = 0.001) and higher levels of childhood trauma (median CTQ score: 33 vs. 28, P = 0.002), but received less social support (median SSRS score: 37 vs. 45, P 0.05). Table 1 Comparison of baseline characteristics between patients with DD and HCs HC group (n = 41) DD group (n = 69) P Age (years) 52 (49-57) 54 (46-60) 0.251 Female (n, %) 31 (75.6) 47 (68.1) 0.403 Educational level 0.391 Elementary school or lower 7 (17.1) 19 (27.5) Secondary school or vocational school 24 (58.5) 38 (55.1) College or higher 10 (24.4) 12 (17.4) Marriage status 0.538 Married 38 (92.7) 60 (87.0) Single, divorced or widowed 3 (7.3) 9 (13.0) BMI (kg/m2) 23.66±2.63 21.83±2.94 0.001 HAMD-21 1 (0-2) 22 (17-24) < 0.001 HAMA 1 (0-2) 22 (20-27) < 0.001 PHQ-15 2 (0-4) 12 (10-14) < 0.001 PSQI 4 (2-6) 16 (13-17) < 0.001 CPSS 26 (23-32) 40 (32-48) < 0.001 TEPS 86 (74-92) 72 (59-82) < 0.001 CTQ 28 (26-32) 33 (27-44) 0.002 SSRS 45 (42-49) 37 (33-40) < 0.001 DSST 40 (31-53) 36 (25-54) 0.443 Note. DD = depressive disorder; HC = healthy control; BMI = body mass index; HAMD-21 = 21-item Hamilton Depression Rating Scale; HAMA = Hamilton Anxiety Rating Scale; PHQ-15 = Patient Health Questionnaire-15; PSQI = Pittsburgh Sleep Quality Index; CPSS = Chinese Perceived Stress Scale; TEPS = Temporal Experience of Pleasure Scale; CTQ = Childhood Trauma Questionnaire; SSRS = Social Support Rate Scale; DSST = Digit Symbol Substitution Test. Microstate Characteristics Figure 2 depicts the spatial topographies of the four microstates (A, B, C, and D) for both the DD and HC groups. TANOVA analysis using Ragu software revealed no significant differences between the two groups ( P > 0.05). As detailed in Supplemental Table S1 and Figure 3, the DD group exhibited a shorter median duration of microstate C (75.87ms vs. 80.99ms, P = 0.017), and a higher median transition probability from microstate D to B (0.34 vs. 0.30, P = 0.002) compared to the HC group. However, there were no significant differences in microstate coverage or occurrence between the groups ( P > 0.05). After FDR correction, the difference in transition probability from microstate D to B remained significant ( P FDR = 0.048), whereas the difference in the duration of microstate C was no longer significant ( P FDR = 0.204). Functional Connectivity Features According to Lee JY (2022)[35], the following brain regions were selected for FC analysis between the two groups: the default mode network (DMN) included bilateral BA2, BA7, BA10, BA11, BA19, BA29, BA30, BA31, BA35, and BA39; the executive control network (ECN) included bilateral BA40, BA44, and BA46; the salience network (SN) included bilateral amygdala, BA13, BA22, BA23, and BA24. As shown in Figure 4, compared to the HCs, DD patients exhibited widespread reductions in FC and localized increases in FC. Specifically, increased connectivity was observed in the following regions: 1) In the δ band, primarily within certain areas of the DMN; 2) In the θ band, within specific regions of the DMN and between the DMN and SN; 3) In the α band, within parts of the SN and between the DMN and SN; and 4) In the β band, within the DMN. However, all these differences have not survived for FDR correction ( P FDR > 0.05). SVM results The above results indicate distinct alterations in microstates and FC in patients with DD. To determine whether combing microstate and FC features can better distinguish patients with DD from HCs, the data were divided into three feature sets. As shown in Supplemental Table S2 and Figure 5, feature set 3 achieved the best classification performance with the SVM classifier (accuracy = 0.810, sensitivity = 0.923, specificity = 0.625, AUC = 0.837). Discussion This study represents a significant advancement in the diagnosis of DD by integrating microstate and FC analyses using high-density EEG and source localization techniques. While previous studies have examined these features independently, this research is the first to comprehensively combine them, demonstrating their synergistic diagnostic value. With a classification accuracy of 81.0% using a SVM classifier, this study underscores the potential of EEG-based biomarkers as accessible and objective tools for mental health diagnostics. One of the key findings of this study is the significant increase in the transition probability from microstate D to B in the DD group. The finding aligns with previous research reporting alterations in microstate dynamics in patients with DD[ 36 ]. For instance, a previous study examining 47 first-episode drug-naïve adolescents with DD and suicidal ideation, compared to 26 depressed adolescents without suicidal ideation, also reported heightened transition probabilities in microstate D to B in those with suicidal ideation[ 36 ]. In a comparative analysis of microstate characteristics between 63 DD patients and 79 HCs, Murphy et al. reported a significant reduction in the coverage, duration and occurrence of microstate D in the DD group[ 37 ]. Furthermore, in this group, the duration and coverage of microstate D were significantly negatively correlated with symptom severity[ 37 ]. Microstate D, associated with the dorsal attention network (DAN), reflects attentional shifts and reorientation processes[ 15 ], while microstate B, linked to the visual network, mediates self-referential cognitive processing[ 15 , 38 ]. The heightened transition probability suggests instability in maintaining attentional focus and an increased tendency to shift toward self-referential thought processes, which are characteristic of depressive symptoms[ 39 ]. These findings align with previous research suggest that microstate transitions may serve as a potential biomarker for affective-cognitive dysfunctions in DD. In addition, this study found that the duration of microstate C was significantly shorter in patients with DD compared to HCs, though this difference did not survive FDR correction. Microstate C is associated with the ECN and the DMN, which are involved in cognitive control and external evaluation processes[ 40 ]. Abnormalities in this microstate may indicate impairments in these functions among patients with DD[ 40 ]. However, inconsistencies with prior studies reporting increased microstate C duration in different DD populations may be attributable to variations in study populations, age demographics, clustering algorithms, or small sample sizes[ 6 , 19 ]. Future research with standardized methodologies and larger cohorts is necessary to validate these findings and reconcile these discrepancies. This study also identified abnormal FC patterns in patients with DD, characterized by widespread decreases and localized increases, particularly within specific regions of the DMN in the δ, θ, and β frequency bands, and between the DMN and SN in the θ and α bands. These findings are consistent with results from fMRI studies that have predominantly explored FC abnormalities in DD[ 41 , 42 ]. However, unlike fMRI, our EEG-based approach offers the advantage of capturing rapid, frequency-specific network dynamics, providing a fresh perspective on the neural oscillatory disruptions underlying DD[ 4 ]. Although the differences did not survive FDR correction, the observed trends align with prior reports of DMN hyperactivity and SN disruptions, which contribute to negative self-referential thinking and impaired emotion regulation[ 41 , 42 ]. Moreover, this study extends existing EEG research by highlighting frequency-specific disruptions, such as reduced alpha-band connectivity and altered theta-band oscillations, both of which are central to DD pathophysiology[ 43 – 45 ]. The use of EEG to derive FC measures underscores its potential to complement traditional neuroimaging methods by providing a portable and temporally precise tool for investigating network dysfunctions in DD. Microstates have been successfully used in classifying psychiatric disorders. For instance, Baradits et al. achieved an accuracy of 82.7% in differentiating patients with schizophrenia from HCs using SVM classification based on microstate metrics[ 46 ]. Our results indicate that integrating EEG microstate and FC features enhances the accuracy of distinguishing DD patients from HCs, underscoring their value as diagnostic markers for depression. This comprehensive approach addresses a key limitation of earlier studies, which often focused on a single modality. Additionally, the use of source localization for FC analysis enhances the spatial specificity of connectivity measures, reducing the impact of volume conduction and providing more precise insights into network-level abnormalities. By leveraging machine learning, this study sets a precedent for using EEG biomarkers in the clinical diagnosis of psychiatric disorders, particularly for scalable and cost-effective applications. This study has several key strengths. First, while previous research on DD has predominantly focused on MRI, this study employs high-density EEG and source localization techniques to extract both microstate and FC features from DD patients and HCs, offering a more holistic perspective on the pathophysiological mechanisms underlying DD. Second, the use of high-density EEG allowing for more accurate detection of subtle neural activity differences between DD patients and healthy controls. Additionally, source localization reduces the impact of volume conduction and enhances the anatomical specificity of functional connectivity measurements, enabling a more precise mapping of brain network disruptions in DD. This approach bridges the gap between the temporal precision of EEG and the spatial accuracy traditionally associated with neuroimaging modalities like fMRI, offering a comprehensive understanding of DD's pathophysiological mechanisms. Finally, unlike previous studies that concentrated on single feature sets for machine learning, we developed and tested three distinct feature sets and demonstrated that combining microstate and FC features yields superior diagnostic accuracy. Despite these strengths, this study also has certain limitations. First, the relatively small sample size, coupled with an imbalance between DD patients and healthy controls, may limit the generalizability of the findings and could introduce bias into the classification model. Second, the cross-sectional design precludes conclusions about causality or the progression of EEG abnormalities in DD, highlighting the need for longitudinal studies to explore these biomarkers over time. Third, while FDR correction was applied to minimize false positives, it may have reduced statistical power, potentially excluding subtle but meaningful differences. Finally, the lack of external validation limits the generalizability of the SVM classifier, future studies with larger sample sizes and standardized methodologies are needed to validate and extend these findings. Conclusions In conclusion, this study advances the field by integrating microstate and FC features through high-density EEG and source localization techniques. By leveraging machine learning, we demonstrate the diagnostic potential of these combined features, which surpasses the performance of either feature set alone. These findings not only deepen our understanding of DD’s neural mechanisms but also establish a scalable framework for objective, accessible diagnostic tools in mental health care. Future research should aim to validate these findings in larger, diverse populations to maximize their clinical utility. Abbreviations DD Depression disorder WHO World Health Organization EEG electroencephalogram fMRI functional magnetic resonance imaging PET positron emission tomography MEG magnetoencephalography FC functional connectivity DSM-V Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition HAMD-21 Hamilton Depression Rating Scale-21 ICD-10 International Classification of Diseases, Tenth Revision BMI Body Mass Index HAMA Hamilton Anxiety Rating Scale CTQ Childhood Trauma Questionnaire SSRS Social Support Rate Scale PHQ-15 Patient Health Questionnaire-15 PSQI Pittsburgh Sleep Quality Index CPSS Chinese Perceived Stress Scale TEPS Temporal Experience of Pleasure Scale DSST Digit Symbol Substitution Test IIR infinite impulse response ICA independent component analysis GFP global field power LORETA low-resolution electromagnetic tomography analysis ROIs Regions of interest BA Brodmann areas PLI Phase Lag Index SDs standard deviations IQRs interquartile ranges FDR false discovery rate TANOVA topographic analysis of variance SVM support vector machine ROC receiver operating characteristic AUC area under the curve DMN default mode network ECN executive control network SN salience network DAN dorsal attention network. Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (TJ-IRB20220205). The study was registered on http://www.chictr.org.cn/ and the registration number was ChiCTR2200057365 (registration date: March 9, 2022). All methods were performed in accordance with the Declaration of Helsinki. All participants signed informed consent to participate. Consent for publication All participants signed informed content for publication. Availability of data and materials The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. Competing interest The authors report there are no conflict of interest to declare. Funding The study was supported by the National Natural Science Foundation of China (82090034). Authors' contributions YY, HZ, and KW were responsible for the conceptualization and design of this research. CL, KS, and JF collaborated in the data collection for DD patients, while CL, YHS, YX, and ZWW contributed to data collection for HCs. CL conducted the data analysis and wrote the initial manuscript. HZ and YY provided comprehensive revisions to the manuscript. All authors contributed in the manuscript revisions and approved the final version for submission. Acknowledgements The authors would like to thank the participants and their family for participation and referring physicians. References Malhi GS, Mann JJ, Depression. Lancet. 2018;392(10161):2299–312. Available online. https://www.who.int/news-room/fact-sheets/detail/depression Belmaker RH, Agam G. Major depressive disorder. N Engl J Med. 2008;358(1):55–68. de Aguiar Neto FS, Rosa JLG. Depression biomarkers using non-invasive EEG: A review. Neurosci Biobehav Rev. 2019;105:83–93. Biomarkers Definitions Working G. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69(3):89–95. Peng RJ, Fan Y, Li J, Zhu F, Tian Q, Zhang XB. Abnormalities of electroencephalography microstates in patients with depression and their association with cognitive function. World J Psychiatry. 2024;14(1):128–40. Yasin S, Othmani A, Raza I, Hussain SA. Machine learning based approaches for clinical and non-clinical depression recognition and depression relapse prediction using audiovisual and EEG modalities: A comprehensive review. Comput Biol Med. 2023;159:106741. Parsa M, Rad HY, Vaezi H, Hossein-Zadeh GA, Setarehdan SK, Rostami R, et al. EEG-based classification of individuals with neuropsychiatric disorders using deep neural networks: A systematic review of current status and future directions. Comput Methods Programs Biomed. 2023;240:107683. Klooster D, Voetterl H, Baeken C, Arns M. Evaluating Robustness of Brain Stimulation Biomarkers for Depression: A Systematic Review of Magnetic Resonance Imaging and Electroencephalography Studies. Biol Psychiatry. 2024;95(6):553–63. Liang J, Yu Q, Liu Y, Qiu Y, Tang R, Yan L, et al. Gray matter abnormalities in patients with major depressive disorder and social anxiety disorder: a voxel-based meta-analysis. Brain Imaging Behav. 2023;17(6):749–63. Cha J, Choi KS, Rajendra JK, McGrath CL, Riva-Posse P, Holtzheimer PE, et al. Whole brain network effects of subcallosal cingulate deep brain stimulation for treatment-resistant depression. Mol Psychiatry. 2024;29(1):112–20. Xia Y, Wang X, Sheng J, Hua L, Dai Z, Sun H, et al. Response inhibition related neural oscillatory patterns show reliable early identification of bipolar from unipolar depression in a Go/No-Go task. J Affect Disord. 2024;351:414–24. Khanna A, Pascual-Leone A, Michel CM, Farzan F. Microstates in resting-state EEG: current status and future directions. Neurosci Biobehav Rev. 2015;49:105–13. Miljevic A, Bailey NW, Murphy OW, Perera MPN, Fitzgerald PB. Alterations in EEG functional connectivity in individuals with depression: A systematic review. J Affect Disord. 2023;328:287–302. Britz J, Van De Ville D, Michel CM. BOLD correlates of EEG topography reveal rapid resting-state network dynamics. NeuroImage. 2010;52(4):1162–70. Michel CM, Koenig T. EEG microstates as a tool for studying the temporal dynamics of whole-brain neuronal networks: A review. NeuroImage. 2018;180(Pt B):577–93. Lei L, Liu Z, Zhang Y, Guo M, Liu P, Hu X, et al. EEG microstates as markers of major depressive disorder and predictors of response to SSRIs therapy. Prog Neuropsychopharmacol Biol Psychiatry. 2022;116:110514. Qin X, Xiong J, Cui R, Zou G, Long C, Lei X. EEG microstate temporal Dynamics Predict depressive symptoms in College Students. Brain Topogr. 2022;35(4):481–94. Zhao Z, Niu Y, Zhao X, Zhu Y, Shao Z, Wu X et al. EEG microstate in first-episode drug-naive adolescents with depression. J Neural Eng 2022; 19(5). Tura A, Goya-Maldonado R. Brain connectivity in major depressive disorder: a precision component of treatment modalities? Transl Psychiatry. 2023;13(1):196. Sun S, Yan C, Qu S, Luo G, Liu X, Tian F, et al. Resting-state dynamic functional connectivity in major depressive disorder: A systematic review. Prog Neuropsychopharmacol Biol Psychiatry. 2024;135:111076. Mumtaz W, Ali SSA, Yasin MAM, Malik AS. A machine learning framework involving EEG-based functional connectivity to diagnose major depressive disorder (MDD). Med Biol Eng Comput. 2018;56(2):233–46. Chen H, Lei Y, Li R, Xia X, Cui N, Chen X, et al. Resting-state EEG dynamic functional connectivity distinguishes non-psychotic major depression, psychotic major depression and schizophrenia. Mol Psychiatry. 2024;29(4):1088–98. Gratton G, Coles MG, Donchin E. A new method for off-line removal of ocular artifact. Electroencephalogr Clin Neurophysiol. 1983;55(4):468–84. Nagabhushan Kalburgi S, Kleinert T, Aryan D, Nash K, Schiller B, Koenig T. MICROSTATELAB: The EEGLAB Toolbox for Resting-State Microstate Analysis. Brain Topogr.; 2023. Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9–21. Schoffelen JM, Gross J. Source connectivity analysis with MEG and EEG. Hum Brain Mapp. 2009;30(6):1857–65. Fan J, Li W, Lin M, Li X, Deng X. Effects of mindfulness and fatigue on emotional processing: an event-related potentials study. Front Behav Neurosci. 2023;17:1175067. Fachner JC, Maidhof C, Grocke D, Nygaard Pedersen I, Trondalen G, Tucek G, et al. Telling me not to worry… Hyperscanning and Neural Dynamics of Emotion Processing During Guided Imagery and Music. Front Psychol. 2019;10:1561. Pascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MC, Hell D, et al. Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia. Psychiatry Res. 1999;90(3):169–79. Greber M, Klein C, Leipold S, Sele S, Jancke L. Heterogeneity of EEG resting-state brain networks in absolute pitch. Int J Psychophysiol. 2020;157:11–22. Stam CJ, Nolte G, Daffertshofer A. Phase lag index: assessment of functional connectivity from multi channel EEG and MEG with diminished bias from common sources. Hum Brain Mapp. 2007;28(11):1178–93. Koenig T, Kottlow M, Stein M, Melie-Garcia L. Ragu: a free tool for the analysis of EEG and MEG event-related scalp field data using global randomization statistics. Comput Intell Neurosci. 2011; 2011:938925. Zou H. iHBPs-VWDC: variable-length window-based dynamic connectivity approach for identifying hormone-binding proteins. J Biomol Struct Dyn. 2023:1–10. Lee JY, Choi CH, Park M, Park S, Choi JS. Enhanced resting-state EEG source functional connectivity within the default mode and reward-salience networks in Internet gaming disorder - CORRIGENDUM. Psychol Med. 2022;52(11):2199–200. He XQ, Hu JH, Peng XY, Zhao L, Zhou DD, Ma LL, et al. EEG microstate analysis reveals large-scale brain network alterations in depressed adolescents with suicidal ideation. J Affect Disord. 2024;346:57–63. Murphy M, Whitton AE, Deccy S, Ironside ML, Rutherford A, Beltzer M, et al. Abnormalities in electroencephalographic microstates are state and trait markers of major depressive disorder. Neuropsychopharmacology. 2020;45(12):2030–7. Tarailis P, Koenig T, Michel CM, Griskova-Bulanova I. The Functional Aspects of Resting EEG Microstates: A Systematic Review. Brain Topogr. 2024;37(2):181–217. Norton DJ, McBain RK, Pizzagalli DA, Cronin-Golomb A, Chen Y. Dysregulation of visual motion inhibition in major depression. Psychiatry Res. 2016;240:214–21. Croce P, Zappasodi F, Capotosto P. Offline stimulation of human parietal cortex differently affects resting EEG microstates. Sci Rep. 2018;8(1):1287. Sheline YI, Barch DM, Price JL, Rundle MM, Vaishnavi SN, Snyder AZ, et al. The default mode network and self-referential processes in depression. Proc Natl Acad Sci U S A. 2009;106(6):1942–7. Lemogne C, le Bastard G, Mayberg H, Volle E, Bergouignan L, Lehericy S, et al. In search of the depressive self: extended medial prefrontal network during self-referential processing in major depression. Soc Cogn Affect Neurosci. 2009;4(3):305–12. Mayberg HS. Limbic-cortical dysregulation: a proposed model of depression. J Neuropsychiatry Clin Neurosci. 1997;9(3):471–81. Clancy KJ, Andrzejewski JA, You Y, Rosenberg JT, Ding M, Li W. Transcranial stimulation of alpha oscillations up-regulates the default mode network. Proc Natl Acad Sci U S A 2022; 119(1). Huang Y, Yi Y, Chen Q, Li H, Feng S, Zhou S, et al. Analysis of EEG features and study of automatic classification in first-episode and drug-naive patients with major depressive disorder. BMC Psychiatry. 2023;23(1):832. Baradits M, Bitter I, Czobor P. Multivariate patterns of EEG microstate parameters and their role in the discrimination of patients with schizophrenia from healthy controls. Psychiatry Res. 2020;288:112938. Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 04 May, 2026 Reviews received at journal 04 May, 2026 Reviewers agreed at journal 20 Apr, 2026 Reviewers agreed at journal 15 Apr, 2026 Reviewers agreed at journal 19 May, 2025 Reviewers agreed at journal 12 May, 2025 Reviews received at journal 12 May, 2025 Reviewers agreed at journal 12 May, 2025 Reviewers invited by journal 02 May, 2025 Editor invited by journal 18 Mar, 2025 Editor assigned by journal 18 Mar, 2025 Submission checks completed at journal 18 Mar, 2025 First submitted to journal 06 Mar, 2025 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-6170701","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":451847351,"identity":"ba771c4f-3a03-45e1-a42b-75e1cfb8d345","order_by":0,"name":"Cun Li","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Cun","middleName":"","lastName":"Li","suffix":""},{"id":451847354,"identity":"6a4072da-6e53-4235-a72e-c0ffd4d5b7fd","order_by":1,"name":"Ke Shi","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ke","middleName":"","lastName":"Shi","suffix":""},{"id":451847358,"identity":"14717b11-60c3-412c-82fa-37948ed51866","order_by":2,"name":"Yanhui Song","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Yanhui","middleName":"","lastName":"Song","suffix":""},{"id":451847360,"identity":"29f33782-79f1-47f9-8181-d35bf9d72cfc","order_by":3,"name":"Ye Xia","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ye","middleName":"","lastName":"Xia","suffix":""},{"id":451847362,"identity":"6eba8b63-006f-42a7-880d-a7172855b241","order_by":4,"name":"Ziwei Wang","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ziwei","middleName":"","lastName":"Wang","suffix":""},{"id":451847364,"identity":"d5f260ee-cd72-4ff2-bdaa-743b5b73a1e8","order_by":5,"name":"Jie Feng","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jie","middleName":"","lastName":"Feng","suffix":""},{"id":451847369,"identity":"1a3bbbc5-7911-432e-a27a-3c5304747256","order_by":6,"name":"Kai Wang","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kai","middleName":"","lastName":"Wang","suffix":""},{"id":451847372,"identity":"8130f9d9-a246-4d8d-af17-eab31ced70ce","order_by":7,"name":"Han Zhang","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Han","middleName":"","lastName":"Zhang","suffix":""},{"id":451847375,"identity":"3fe30ed3-706b-4148-816f-8e38a2e8e800","order_by":8,"name":"Yuan Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA90lEQVRIiWNgGAWjYDAD+8PMx8AMNnai9RxvS2NgSABqYSZay5kzZmAtDIS0GNxIfvaYp+aOXeOMnG8PPv7YJs/HzMD44WMOPi1p5sY8x54lN0vkbjeckXDbsI2ZgVly5jZ8WhLMpHnYDiezSeRuk+ZJuM0I1MLGzItXS/o3aZ5/h5N5JHKegbTYE6Elx0yat+2wnQTPGTaQlkSCWiTPvCmTnNt3OMGAvc1Mckba7eQ2ZsZmvH7hO56+TeLNt8P2BszMzyQ+2Ny2nd/efPDDRzxaFA4wMDDxMDAkNiDEGBtwKIYAeaA04w9gesGrahSMglEwCkY2AABl7FC1aJksfwAAAABJRU5ErkJggg==","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Yuan","middleName":"","lastName":"Yang","suffix":""}],"badges":[],"createdAt":"2025-03-06 12:53:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6170701/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6170701/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":82349952,"identity":"4ce17de6-f2be-4d02-b503-23b921a3e387","added_by":"auto","created_at":"2025-05-09 10:49:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1248680,"visible":true,"origin":"","legend":"\u003cp\u003eFlow diagram of individuals from enrollment.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003e DD = depressive disorder; HC = healthy control; EEG = electroencephalogram.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/9b647a70fa5330db60b43013.png"},{"id":82349975,"identity":"f8ab2eb6-292d-414c-93f8-fd8e7b4037e9","added_by":"auto","created_at":"2025-05-09 10:49:46","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":64317057,"visible":true,"origin":"","legend":"\u003cp\u003eTopography of four microstates in the healthy control (HC) group (top) and the depressive disorder (DD) group (bottom).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003e A, B, C and D in the figure represent microstate A, microstate B, microstate C and microstate D, respectively.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/fc7d5b85057a9b520dc18926.png"},{"id":82349959,"identity":"7228a7ef-7f20-40fc-a5d0-9807f8c87053","added_by":"auto","created_at":"2025-05-09 10:49:46","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":4197025,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of microstate characteristics between the healthy control (HC) group and the depressive disorder (DD) group.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003e A, B, C and D in the figure represent microstate A, microstate B, microstate C and microstate D, respectively.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/7846e4a803f257378e569f76.png"},{"id":82349990,"identity":"5731d110-5883-4101-be61-810c4275d39e","added_by":"auto","created_at":"2025-05-09 10:49:49","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":55788151,"visible":true,"origin":"","legend":"\u003cp\u003eRegions of changed functional connectivity (FC) in the healthy control (HC) group and depressive disorder (DD) group in four frequency bands.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003eL= left; R = right.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/6a66a76f7642c596e0e54ad4.png"},{"id":82349955,"identity":"0f1618dd-78e0-4d68-aee4-6729128535b1","added_by":"auto","created_at":"2025-05-09 10:49:46","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":381406,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver operating characteristic (ROC) curves for three feature sets in support vector machine (SVM) classifier\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote. \u003c/em\u003eSPE= specificity; SEN = sensitivity; AUC = area under the curve.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/9e80c90acd3ec7e0e0e9fee7.png"},{"id":82349956,"identity":"f01e8beb-64a6-42c3-b67d-05456993656c","added_by":"auto","created_at":"2025-05-09 10:49:46","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":713615,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-6170701/v1/b57ade526ec4a63adb293ba5.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"EEG Microstates and Functional Connectivity Abnormalities in Depressive Disorders: Diagnostic Potential and Machine Learning Application","fulltext":[{"header":"Introduction","content":"\u003cp\u003eDepression disorder (DD) is a prevalent and debilitating mental health condition characterized by persistent low mood, anhedonia, and a range of cognitive, emotional, and somatic symptoms[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. According to the World Health Organization (WHO), DD affects approximately 3.8% of the global population, with a particularly high prevalence of 5.0% among adults[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Despite its prevalence and impact, diagnosing DD is challenging due to the heavy reliance on clinical interviews and subjective symptom reporting, lacking objective biomarkers[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. This highlights the urgent need to identify reliable and objective markers to enhance diagnostic accuracy and guide personalized treatment strategies.\u003c/p\u003e \u003cp\u003ePrior research has indicated that aberrant brain structure and function are underlying the pathophysiological mechanisms of DD[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. A variety of neuroimaging techniques, including electroencephalogram (EEG) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], functional magnetic resonance imaging (fMRI)[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], positron emission tomography (PET)[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], and magnetoencephalography (MEG)[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], have been widely employed to investigate these mechanisms. Among these methods, EEG has gained considerable attention due to its ability to directly measure neuronal activity, coupled with its high temporal resolution and portability[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Moreover, advanced computational methods, such as microstate analysis and functional connectivity (FC) analysis, have been developed to extract meaningful features from EEG signals, providing novel insights into the dynamic spatiotemporal patterns of brain activity [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMicrostates, which are brief, collections of EEG topographic maps that typically dominate for 80-120ms before transitioning to another map, offer valuable insights into the dynamics of large-scale brain networks[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Research has consistently identified four canonical microstates (A, B, C, and D), which are closely related to specific resting-state brain networks[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Studies have reported abnormal microstate patterns in DD, such as altered durations, occurrences, and coverages of specific microstates[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Moreover, microstate features have demonstrated potential as diagnostic biomarkers for DD when combined with machine learning algorithms[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFC analysis is another widely used approach to investigate the interactions among brain regions and networks[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. While fMRI-based FC studies have provided valuable insights into the network abnormalities in DD, they are limited by the low temporal resolution of fMRI, which typically measures brain activity on a timescale of seconds[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. In contrast, EEG allowing for the investigation of FC at shorter timescales[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], which is particularly important for understanding the rapid fluctuations in brain network interactions that underlie cognitive and affective processes often impaired in DD[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMachine learning and feature selection methods have been utilized to differentiate DD patients from HCs based on FC[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. For example, Chen et al. employed dynamic EEG FC features to classify DD patients with and without psychotic symptoms, schizophrenia patients, and HCs, obtaining an accuracy of 73.1% in the four-group classification[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWhile both microstate and FC features have shown potential in distinguishing DD patients from HCs, prior research has been limited by inconsistent findings, small sample sizes and a lack of comprehensive analysis that combining both microstate and FC for improved classification.\u003c/p\u003e \u003cp\u003eTo address these limitations, the present study set out to test the following hypotheses: (1) DD patients exhibit distinctive EEG microstate features and widespread abnormal brain network connectivity; and (2) The combination of microstate and FC features, analyzed with machine learning algorithms, can more effectively differentiate DD patients from HCs. By utilizing high-density EEG and source localization techniques, this study seeks to enhance the spatial and temporal resolution of EEG-based biomarkers, thereby providing a more comprehensive understanding of DD's pathophysiological mechanisms. Our findings may not only deepen our understanding of the neural mechanisms underlying DD but also facilitate the development of more targeted and effective interventions for this prevalent and debilitating disorder.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eParticipants\u003c/h2\u003e\n \u003cp\u003eThis present study enrolled 115 untreated patients diagnosed with DD from Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology from March 2022 to September 2023. The eligibility criteria for participants were as follows: (1) age range between 18 and 65 years; (2) aligning with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) criteria for diagnosing DD; (3) a Hamilton Depression Rating Scale-21 (HAMD-21) score of 17 or higher; (4) a minimum education level equivalent to elementary school; (5) right-handedness; and (6) native Chinese speaker. The exclusion criteria encompassed: (1) a history of mental illnesses such as schizophrenia, severe DD with suicidal ideation, alcohol dependence, or substance misuse; (2) the presence of significant or unstable medical conditions; (3) any current or past head trauma, central nervous system disorders, or other conditions listed under the International Classification of Diseases, Tenth Revision (ICD-10); (4) contraindications to antidepressant medications; (5) use (within the past month) of antidepressants or long-acting antipsychotic medications; (6) impairments such as aphasia, hearing loss, visual impairment, or cognitive dysfunction; and (7) pregnancy, breastfeeding, or plans to conceive during the study period. Additionally, 43 HCs matched for age, gender, and education level were included. These participants met the same exclusion criteria, with the added requirement of no history of psychiatric disorders. A flowchart illustrating the recruitment process is presented in Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003eInformed consent was obtained following the principles outlined in the Declaration of Helsinki. The study protocol received ethical approval from the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (approval number: TJ-IRB20220205). The study was registered on \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.chictr.org.cn/\u003c/span\u003e\u003c/span\u003e and the registration number was ChiCTR2200057365 (registration date: March 9, 2022).\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eData Collection\u003c/h3\u003e\n\u003cp\u003eFirstly, general demographic data were collected, including age, gender, educational level, marital status, and Body Mass Index (BMI). Educational level was categorized into three tiers: primary school or below, secondary school/vocational school, and college/university degree or higher. Marital status was classified as married or single/divorced/widowed.\u003c/p\u003e\n\u003cp\u003eIn addition, psychological assessment data were collected using the HAMD-21, Hamilton Anxiety Rating Scale (HAMA), Childhood Trauma Questionnaire (CTQ), Social Support Rate Scale (SSRS), Patient Health Questionnaire-15 (PHQ-15), Pittsburgh Sleep Quality Index (PSQI), Chinese Perceived Stress Scale (CPSS), Temporal Experience of Pleasure Scale (TEPS), and Digit Symbol Substitution Test (DSST).\u003c/p\u003e\n\u003ch3\u003eEEG Recording and Preprocessing\u003c/h3\u003e\n\u003cp\u003eRest-state EEG signals were recorded during 12-minute sessions that alternated between eyes-open and eyes-closed conditions. Recordings took place in a dimly lit, electrically shielded, and soundproof chamber using a 128-channel EEG system (BrainVision Recorder software, Brain Products GmbH, Germany). During the eyes-open condition, participants were instructed to stay as still as possible, avoid blinking or making eye movements, and focus on a central fixation cross. The electrodes were arranged following the international 10/5 system, with a sampling rate of 1000 Hz and the FCz electrode serving as the reference. Impedance levels were maintained below 20 KΩ throughout the recording process.\u003c/p\u003e\n\u003cp\u003eEEG data preprocessing was performed using BrainVision Analyzer software (version 2.2, Brain Products GmbH, Germany) and included the following steps: (1) Re-referencing: Raw EEG data were re-referenced offline to the average of the mastoid electrodes (TP9 and TP10). (2) Filtering: Data were filtered using infinite impulse response (IIR) filters (0.5\u0026ndash;45 Hz) and a 50 Hz notch filter. (3) Down sampling: The sampling frequency of rest-state EEG data was reduced to 500Hz. (4) Ocular correction: Eyeblink and ocular movement artifacts were corrected using independent component analysis (ICA) based on established standards[\u003cspan class=\"CitationRef\"\u003e24\u003c/span\u003e]. (5) Artifact rejection: Segments with voltage changes exceeding 50\u0026micro;V between sample points, a voltage difference of 300\u0026micro;V within a segment, or a maximum voltage difference of less than 0.5\u0026micro;V within 100ms were automatically rejected, with interpolation for electrodes with more than 20% bad segments. (6) Segmentation: The cleaned resting-state EEG data was segmented into 2-second trails.\u003c/p\u003e\n\u003ch3\u003eMicrostates Analysis\u003c/h3\u003e\n\u003cp\u003eMicrostate analysis was conducted using the Microstates toolbox[\u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e] integrated within the EEGLAB plugin[\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e] in MATLAB R2023a. First, the global field power (GFP) was calculated for each participant at every time point. Given that the local maxima of the GFP curve have the strongest signal strength and highest signal-to-noise ratio, the corresponding topographical maps were selected for subsequent clustering analysis. Next, a modified k-means algorithm was utilized to cluster the topographical maps, yielding four distinct microstates (A-D) for each group. The group-level microstates were then used as template maps. Each participant\u0026apos;s topographical maps were assigned to one of the four microstate classes (A-D) based on spatial similarity with the templates. Finally, key microstate parameters, including mean duration, occurrence, coverage, transition probability, and topographic shape, were extracted for each participant.\u003c/p\u003e\n\u003ch3\u003eSource Localization Analyses\u003c/h3\u003e\n\u003cp\u003eFC derived from scalp electrodes is susceptible to volume conduction effects, which can produce artificial connections[\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e]. To address this limitation, calculating FC in the source space can greatly improve accuracy[\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e]. In this study, we performed source localization analysis using low-resolution electromagnetic tomography analysis (LORETA) within the BrainVision Analyzer software (version 2.2, Brain Products GmbH, Germany) in the present study[\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e29\u003c/span\u003e]. LORETA estimates the current density (\u0026micro;A/mm\u003csup\u003e2\u003c/sup\u003e) of intracranial sources that generate scalp-recorded EEG signals[\u003cspan class=\"CitationRef\"\u003e30\u003c/span\u003e]. Regions of interest (ROIs) were defined based on a publicly available source file (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.brainproducts.com/files/public/downloads/LORETA/All_ROIs.xml\u003c/span\u003e\u003c/span\u003e). The source space encompasses 88 regions, including the bilateral amygdala, bilateral Brodmann areas (BA) (BA1-11, BA13, BA17-25, BA27-47), and bilateral hippocampus[\u003cspan class=\"CitationRef\"\u003e31\u003c/span\u003e], as illustrated in Supplemental Figure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003eFunctional Connectivity\u003c/h2\u003e\n \u003cp\u003eIn this study, the Phase Lag Index (PLI) was chosen to assess the strength of FC between different ROIs[\u003cspan class=\"CitationRef\"\u003e32\u003c/span\u003e]. PLI, a phase-based connectivity analysis method, is predicted on the distribution of phase angle differences between electrodes pairs[\u003cspan class=\"CitationRef\"\u003e32\u003c/span\u003e]. The underlying principle is that when neural clusters are functionally coupled, their oscillatory sequences become phase-synchronized.\u003c/p\u003e\n \u003cp\u003eFor a more comprehensive understanding of brain connectivity, EEG signals were segmented into four frequency sub-bands: delta (\u0026delta;, 1\u0026ndash;4 Hz), theta (\u0026theta;, 4\u0026ndash;8 Hz), alpha (\u0026alpha;, 8\u0026ndash;12 Hz), and beta (\u0026beta;, 13\u0026ndash;30 Hz). The FC values between each pair of ROIs were subsequently calculated across these frequency bands.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003eStatistical Analysis\u003c/h2\u003e\n \u003cp\u003eThe general characteristics of the study population were summarized as follows: as categorical variables were presented as percentages, continuous variables with normal distributions were reported as means\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviations (SDs), and continuous variables with non-normal distributions were described using medians (interquartile ranges, IQRs). Comparisons between the DD and HC groups were performed using chi-square tests or Fisher\u0026apos;s exact tests for categorical variables. For continuous variables, independent two-sample t-tests were employed for normally distributed data, and the Mann-Whitney U test for non-normally distributed data. All statistical analyses were conducted using SPSS (IBM, version 26.0), with the false discovery rate (FDR) correction applied to control for multiple comparisons. Statistical significance was determined using two-tailed tests with a significance level of \u0026alpha;\u0026thinsp;=\u0026thinsp;0.05.\u003c/p\u003e\n \u003cp\u003eAdditionally, topographic analysis of variance (TANOVA) based on Rugu software was implemented to explore differences in microstate topographies between the DD and HC groups. TANOVA is a robust, hypothesis-free randomization test that facilitates the statistical comparison of EEG scalp field topographies across two or more conditions[\u003cspan class=\"CitationRef\"\u003e33\u003c/span\u003e].\u003c/p\u003e\n \u003cp\u003eFurthermore, support vector machine (SVM) algorithms were employed to assess the accuracy of microstate and FC features in distinguishing DD patients from HCs. The steps were as follows: (1) Feature matrix construction: For each frequency band, the FC matrix (88\u0026times;88) was symmetric; hence, the diagonal elements were removed, and the upper triangular elements of the matrix were extracted as classification features, resulting in 88 \u0026times; (88\u0026thinsp;\u0026minus;\u0026thinsp;1)/2 dimensions for each frequency band. (2) Feature selection: F-score values were utilized to evaluate the correlation of each feature with the disease classification, allowing for dimensionality reduction by retaining the most discriminative features while discarding less informative ones[\u003cspan class=\"CitationRef\"\u003e34\u003c/span\u003e]. (3) Feature set construction: Feature set 1 comprised microstate features (33 dimensions); feature set 2 included FC features (15,312 dimensions); and feature set 3 included both microstate and FC features (15,345 dimensions). (4) Training and testing set division: The stratified holdout method was applied, dividing the data into a training set (80%) and a testing set (20%); 5) SVM model: The SVM algorithm was implemented to address the binary classification problem; 6) Model evaluation: The classifier\u0026apos;s generalization ability was quantified using metrics such as accuracy, sensitivity, specificity, receiver operating characteristic (ROC) curves, and the area under the curve (AUC).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eStudy participants\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study enrolled 115 DD patients and 43 HCs from March 2022 to September 2023. After excluding DD patients with missing baseline EEG data (n = 6), a history of antidepressant use (n = 23), and poor-quality EEG data (n = 22), along with two HCs due two poor-quality EEG data, 69 DD patients and 41 HCs were included in the final analysis.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThe participant selection process is shown in Figure 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBaseline characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs presented in Table 1, the DD group had a significantly lower BMI (21.83 \u0026plusmn; 2.94 kg/m\u0026sup2; vs. 23.66 \u0026plusmn; 2.63 kg/m\u0026sup2;, \u003cem\u003eP\u003c/em\u003e = 0.001) and higher levels of childhood trauma (median CTQ score: 33 vs. 28, \u003cem\u003eP\u003c/em\u003e = 0.002), but received less social support (median SSRS score: 37 vs. 45, \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001) compared to the HC group. No significant differences were observed between the groups in terms of age, gender, or educational level (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1\u003c/strong\u003e Comparison of baseline characteristics between patients with DD and HCs\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003eHC group (n = 41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003eDD group (n = 69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eAge (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e52 (49-57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e54 (46-60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.251\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eFemale (n, %)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e31 (75.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e47 (68.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.403\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eEducational level\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.391\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eElementary school or lower\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e7 (17.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e19 (27.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eSecondary school or vocational school\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e24 (58.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e38 (55.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eCollege or higher\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e10 (24.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e12 (17.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eMarriage status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.538\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eMarried\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e38 (92.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e60 (87.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 245px;\"\u003e\n \u003cp\u003eSingle, divorced or widowed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e3 (7.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e9 (13.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eBMI (kg/m2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e23.66\u0026plusmn;2.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e21.83\u0026plusmn;2.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eHAMD-21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e1 (0-2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e22 (17-24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eHAMA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e1 (0-2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e22 (20-27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003ePHQ-15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e2 (0-4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e12 (10-14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003ePSQI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e4 (2-6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e16 (13-17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eCPSS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e26 (23-32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e40 (32-48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eTEPS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e86 (74-92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e72 (59-82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eCTQ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e28 (26-32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e33 (27-44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eSSRS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e45 (42-49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e37 (33-40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 245px;\"\u003e\n \u003cp\u003eDSST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e40 (31-53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 123px;\"\u003e\n \u003cp\u003e36 (25-54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 62px;\"\u003e\n \u003cp\u003e0.443\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003e DD = depressive disorder; HC = healthy control; BMI = body mass index; HAMD-21 = 21-item Hamilton Depression Rating Scale;\u0026nbsp;HAMA = Hamilton Anxiety Rating Scale; PHQ-15 = Patient Health Questionnaire-15; PSQI = Pittsburgh Sleep Quality Index; CPSS = Chinese Perceived Stress Scale; TEPS = Temporal Experience of Pleasure Scale; CTQ = Childhood Trauma Questionnaire; SSRS = Social Support Rate Scale; DSST =\u0026nbsp;Digit Symbol Substitution Test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMicrostate Characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure 2 depicts the spatial topographies of the four microstates (A, B, C, and D) for both the DD and HC groups. TANOVA analysis using Ragu software revealed no significant differences between the two groups (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05). As detailed in Supplemental Table S1 and Figure 3, the DD group exhibited a shorter median duration of microstate C (75.87ms vs. 80.99ms, \u003cem\u003eP\u003c/em\u003e = 0.017), and a higher median transition probability from microstate D to B (0.34 vs. 0.30, \u003cem\u003eP\u003c/em\u003e = 0.002) compared to the HC group. However, there were no significant differences in microstate coverage or occurrence between the groups (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05). After FDR correction, the difference in transition probability from microstate D to B remained significant (\u003cem\u003eP\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e = 0.048), whereas the difference in the duration of microstate C was no longer significant (\u003cem\u003eP\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e = 0.204).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunctional Connectivity Features\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccording to Lee JY (2022)[35], the following brain regions were selected for FC analysis between the two groups: the default mode network (DMN) included bilateral BA2, BA7, BA10, BA11, BA19, BA29, BA30, BA31, BA35, and BA39; the executive control network (ECN) included bilateral BA40, BA44, and BA46; the salience network (SN) included bilateral amygdala, BA13, BA22, BA23, and BA24.\u003c/p\u003e\n\u003cp\u003eAs shown in Figure 4, compared to the HCs, DD patients exhibited widespread reductions in FC and localized increases in FC. Specifically, increased connectivity was observed in the following regions: 1) In the \u0026delta; band, primarily within certain areas of the DMN; 2) In the \u0026theta; band, within specific regions of the DMN and between the DMN and SN; 3) In the \u0026alpha; band, within parts of the SN and between the DMN and SN; and 4) In the \u0026beta; band, within the DMN. However, all these differences have not survived for FDR correction (\u003cem\u003eP\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSVM results\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe above results indicate distinct alterations in microstates and FC in patients with DD. To determine whether combing microstate and FC features can better distinguish patients with DD from HCs, the data were divided into three feature sets. As shown in Supplemental Table S2 and Figure 5, feature set 3 achieved the best classification performance with the SVM classifier (accuracy = 0.810, sensitivity = 0.923, specificity = 0.625, AUC = 0.837).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study represents a significant advancement in the diagnosis of DD by integrating microstate and FC analyses using high-density EEG and source localization techniques. While previous studies have examined these features independently, this research is the first to comprehensively combine them, demonstrating their synergistic diagnostic value. With a classification accuracy of 81.0% using a SVM classifier, this study underscores the potential of EEG-based biomarkers as accessible and objective tools for mental health diagnostics.\u003c/p\u003e \u003cp\u003eOne of the key findings of this study is the significant increase in the transition probability from microstate D to B in the DD group. The finding aligns with previous research reporting alterations in microstate dynamics in patients with DD[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. For instance, a previous study examining 47 first-episode drug-na\u0026iuml;ve adolescents with DD and suicidal ideation, compared to 26 depressed adolescents without suicidal ideation, also reported heightened transition probabilities in microstate D to B in those with suicidal ideation[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. In a comparative analysis of microstate characteristics between 63 DD patients and 79 HCs, Murphy et al. reported a significant reduction in the coverage, duration and occurrence of microstate D in the DD group[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Furthermore, in this group, the duration and coverage of microstate D were significantly negatively correlated with symptom severity[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMicrostate D, associated with the dorsal attention network (DAN), reflects attentional shifts and reorientation processes[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], while microstate B, linked to the visual network, mediates self-referential cognitive processing[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. The heightened transition probability suggests instability in maintaining attentional focus and an increased tendency to shift toward self-referential thought processes, which are characteristic of depressive symptoms[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. These findings align with previous research suggest that microstate transitions may serve as a potential biomarker for affective-cognitive dysfunctions in DD.\u003c/p\u003e \u003cp\u003eIn addition, this study found that the duration of microstate C was significantly shorter in patients with DD compared to HCs, though this difference did not survive FDR correction. Microstate C is associated with the ECN and the DMN, which are involved in cognitive control and external evaluation processes[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Abnormalities in this microstate may indicate impairments in these functions among patients with DD[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. However, inconsistencies with prior studies reporting increased microstate C duration in different DD populations may be attributable to variations in study populations, age demographics, clustering algorithms, or small sample sizes[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Future research with standardized methodologies and larger cohorts is necessary to validate these findings and reconcile these discrepancies.\u003c/p\u003e \u003cp\u003eThis study also identified abnormal FC patterns in patients with DD, characterized by widespread decreases and localized increases, particularly within specific regions of the DMN in the δ, θ, and β frequency bands, and between the DMN and SN in the θ and α bands. These findings are consistent with results from fMRI studies that have predominantly explored FC abnormalities in DD[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. However, unlike fMRI, our EEG-based approach offers the advantage of capturing rapid, frequency-specific network dynamics, providing a fresh perspective on the neural oscillatory disruptions underlying DD[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Although the differences did not survive FDR correction, the observed trends align with prior reports of DMN hyperactivity and SN disruptions, which contribute to negative self-referential thinking and impaired emotion regulation[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Moreover, this study extends existing EEG research by highlighting frequency-specific disruptions, such as reduced alpha-band connectivity and altered theta-band oscillations, both of which are central to DD pathophysiology[\u003cspan additionalcitationids=\"CR44\" citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. The use of EEG to derive FC measures underscores its potential to complement traditional neuroimaging methods by providing a portable and temporally precise tool for investigating network dysfunctions in DD.\u003c/p\u003e \u003cp\u003eMicrostates have been successfully used in classifying psychiatric disorders. For instance, Baradits et al. achieved an accuracy of 82.7% in differentiating patients with schizophrenia from HCs using SVM classification based on microstate metrics[\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Our results indicate that integrating EEG microstate and FC features enhances the accuracy of distinguishing DD patients from HCs, underscoring their value as diagnostic markers for depression. This comprehensive approach addresses a key limitation of earlier studies, which often focused on a single modality. Additionally, the use of source localization for FC analysis enhances the spatial specificity of connectivity measures, reducing the impact of volume conduction and providing more precise insights into network-level abnormalities. By leveraging machine learning, this study sets a precedent for using EEG biomarkers in the clinical diagnosis of psychiatric disorders, particularly for scalable and cost-effective applications.\u003c/p\u003e \u003cp\u003eThis study has several key strengths. First, while previous research on DD has predominantly focused on MRI, this study employs high-density EEG and source localization techniques to extract both microstate and FC features from DD patients and HCs, offering a more holistic perspective on the pathophysiological mechanisms underlying DD. Second, the use of high-density EEG allowing for more accurate detection of subtle neural activity differences between DD patients and healthy controls. Additionally, source localization reduces the impact of volume conduction and enhances the anatomical specificity of functional connectivity measurements, enabling a more precise mapping of brain network disruptions in DD. This approach bridges the gap between the temporal precision of EEG and the spatial accuracy traditionally associated with neuroimaging modalities like fMRI, offering a comprehensive understanding of DD's pathophysiological mechanisms. Finally, unlike previous studies that concentrated on single feature sets for machine learning, we developed and tested three distinct feature sets and demonstrated that combining microstate and FC features yields superior diagnostic accuracy.\u003c/p\u003e \u003cp\u003eDespite these strengths, this study also has certain limitations. First, the relatively small sample size, coupled with an imbalance between DD patients and healthy controls, may limit the generalizability of the findings and could introduce bias into the classification model. Second, the cross-sectional design precludes conclusions about causality or the progression of EEG abnormalities in DD, highlighting the need for longitudinal studies to explore these biomarkers over time. Third, while FDR correction was applied to minimize false positives, it may have reduced statistical power, potentially excluding subtle but meaningful differences. Finally, the lack of external validation limits the generalizability of the SVM classifier, future studies with larger sample sizes and standardized methodologies are needed to validate and extend these findings.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, this study advances the field by integrating microstate and FC features through high-density EEG and source localization techniques. By leveraging machine learning, we demonstrate the diagnostic potential of these combined features, which surpasses the performance of either feature set alone. These findings not only deepen our understanding of DD\u0026rsquo;s neural mechanisms but also establish a scalable framework for objective, accessible diagnostic tools in mental health care. Future research should aim to validate these findings in larger, diverse populations to maximize their clinical utility.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDepression disorder\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eWHO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eWorld Health Organization\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eEEG\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eelectroencephalogram\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003efMRI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003efunctional magnetic resonance imaging\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePET\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003epositron emission tomography\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMEG\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003emagnetoencephalography\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003efunctional connectivity\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDSM-V\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDiagnostic and Statistical Manual of Mental Disorders, Fifth Edition\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHAMD-21\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eHamilton Depression Rating Scale-21\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eICD-10\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eInternational Classification of Diseases, Tenth Revision\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBody Mass Index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHAMA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eHamilton Anxiety Rating Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCTQ\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eChildhood Trauma Questionnaire\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSSRS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSocial Support Rate Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePHQ-15\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePatient Health Questionnaire-15\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePSQI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePittsburgh Sleep Quality Index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCPSS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eChinese Perceived Stress Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTEPS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTemporal Experience of Pleasure Scale\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDSST\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDigit Symbol Substitution Test\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIIR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003einfinite impulse response\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eICA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eindependent component analysis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGFP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eglobal field power\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLORETA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003elow-resolution electromagnetic tomography analysis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eROIs\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRegions of interest\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBrodmann areas\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePLI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePhase Lag Index\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSDs\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003estandard deviations\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eIQRs\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003einterquartile ranges\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eFDR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003efalse discovery rate\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTANOVA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003etopographic analysis of variance\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSVM\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003esupport vector machine\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eROC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ereceiver operating characteristic\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003earea under the curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDMN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edefault mode network\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eECN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eexecutive control network\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003esalience network\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDAN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edorsal attention network.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch4\u003eEthics approval and consent to participate\u003c/h4\u003e\n\u003cp\u003eThis study was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (TJ-IRB20220205). The study was registered on http://www.chictr.org.cn/ and the registration number was ChiCTR2200057365 (registration date: March 9, 2022). All methods were performed in accordance with the Declaration of Helsinki. All participants signed informed consent to participate.\u003c/p\u003e\n\u003ch4\u003eConsent for publication\u003c/h4\u003e\n\u003cp\u003eAll participants signed informed content for publication.\u003c/p\u003e\n\u003ch4\u003eAvailability of data and materials\u003c/h4\u003e\n\u003cp\u003eThe data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.\u003c/p\u003e\n\u003ch4\u003eCompeting interest\u003c/h4\u003e\n\u003cp\u003eThe authors report there are no conflict of interest to declare.\u003c/p\u003e\n\u003ch4\u003eFunding\u003c/h4\u003e\n\u003cp\u003eThe study was supported by the National Natural Science Foundation of China (82090034).\u003c/p\u003e\n\u003ch4\u003eAuthors\u0026apos; contributions\u003c/h4\u003e\n\u003cp\u003eYY, HZ, and KW were responsible for the conceptualization and design of this research. CL, KS, and JF collaborated in the data collection for DD patients, while CL, YHS, YX, and ZWW contributed to data collection for HCs. CL conducted the data analysis and wrote the initial manuscript. HZ and YY provided comprehensive revisions to the manuscript. All authors contributed in the manuscript revisions and approved the final version for submission.\u003c/p\u003e\n\u003ch4\u003eAcknowledgements\u003c/h4\u003e\n\u003cp\u003eThe authors would like to thank the participants and their family for participation and referring physicians.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eMalhi GS, Mann JJ, Depression. Lancet. 2018;392(10161):2299\u0026ndash;312.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAvailable online. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.who.int/news-room/fact-sheets/detail/depression\u003c/span\u003e\u003cspan address=\"https://www.who.int/news-room/fact-sheets/detail/depression\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBelmaker RH, Agam G. Major depressive disorder. N Engl J Med. 2008;358(1):55\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Aguiar Neto FS, Rosa JLG. Depression biomarkers using non-invasive EEG: A review. Neurosci Biobehav Rev. 2019;105:83\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBiomarkers Definitions Working G. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69(3):89\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeng RJ, Fan Y, Li J, Zhu F, Tian Q, Zhang XB. Abnormalities of electroencephalography microstates in patients with depression and their association with cognitive function. World J Psychiatry. 2024;14(1):128\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYasin S, Othmani A, Raza I, Hussain SA. Machine learning based approaches for clinical and non-clinical depression recognition and depression relapse prediction using audiovisual and EEG modalities: A comprehensive review. Comput Biol Med. 2023;159:106741.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParsa M, Rad HY, Vaezi H, Hossein-Zadeh GA, Setarehdan SK, Rostami R, et al. EEG-based classification of individuals with neuropsychiatric disorders using deep neural networks: A systematic review of current status and future directions. Comput Methods Programs Biomed. 2023;240:107683.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKlooster D, Voetterl H, Baeken C, Arns M. Evaluating Robustness of Brain Stimulation Biomarkers for Depression: A Systematic Review of Magnetic Resonance Imaging and Electroencephalography Studies. Biol Psychiatry. 2024;95(6):553\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiang J, Yu Q, Liu Y, Qiu Y, Tang R, Yan L, et al. Gray matter abnormalities in patients with major depressive disorder and social anxiety disorder: a voxel-based meta-analysis. Brain Imaging Behav. 2023;17(6):749\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCha J, Choi KS, Rajendra JK, McGrath CL, Riva-Posse P, Holtzheimer PE, et al. Whole brain network effects of subcallosal cingulate deep brain stimulation for treatment-resistant depression. Mol Psychiatry. 2024;29(1):112\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXia Y, Wang X, Sheng J, Hua L, Dai Z, Sun H, et al. Response inhibition related neural oscillatory patterns show reliable early identification of bipolar from unipolar depression in a Go/No-Go task. J Affect Disord. 2024;351:414\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhanna A, Pascual-Leone A, Michel CM, Farzan F. Microstates in resting-state EEG: current status and future directions. Neurosci Biobehav Rev. 2015;49:105\u0026ndash;13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiljevic A, Bailey NW, Murphy OW, Perera MPN, Fitzgerald PB. Alterations in EEG functional connectivity in individuals with depression: A systematic review. J Affect Disord. 2023;328:287\u0026ndash;302.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBritz J, Van De Ville D, Michel CM. BOLD correlates of EEG topography reveal rapid resting-state network dynamics. NeuroImage. 2010;52(4):1162\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMichel CM, Koenig T. EEG microstates as a tool for studying the temporal dynamics of whole-brain neuronal networks: A review. NeuroImage. 2018;180(Pt B):577\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLei L, Liu Z, Zhang Y, Guo M, Liu P, Hu X, et al. EEG microstates as markers of major depressive disorder and predictors of response to SSRIs therapy. Prog Neuropsychopharmacol Biol Psychiatry. 2022;116:110514.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQin X, Xiong J, Cui R, Zou G, Long C, Lei X. EEG microstate temporal Dynamics Predict depressive symptoms in College Students. Brain Topogr. 2022;35(4):481\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Z, Niu Y, Zhao X, Zhu Y, Shao Z, Wu X et al. EEG microstate in first-episode drug-naive adolescents with depression. J Neural Eng 2022; 19(5).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTura A, Goya-Maldonado R. Brain connectivity in major depressive disorder: a precision component of treatment modalities? Transl Psychiatry. 2023;13(1):196.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun S, Yan C, Qu S, Luo G, Liu X, Tian F, et al. Resting-state dynamic functional connectivity in major depressive disorder: A systematic review. Prog Neuropsychopharmacol Biol Psychiatry. 2024;135:111076.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMumtaz W, Ali SSA, Yasin MAM, Malik AS. A machine learning framework involving EEG-based functional connectivity to diagnose major depressive disorder (MDD). Med Biol Eng Comput. 2018;56(2):233\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen H, Lei Y, Li R, Xia X, Cui N, Chen X, et al. Resting-state EEG dynamic functional connectivity distinguishes non-psychotic major depression, psychotic major depression and schizophrenia. Mol Psychiatry. 2024;29(4):1088\u0026ndash;98.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGratton G, Coles MG, Donchin E. A new method for off-line removal of ocular artifact. Electroencephalogr Clin Neurophysiol. 1983;55(4):468\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNagabhushan Kalburgi S, Kleinert T, Aryan D, Nash K, Schiller B, Koenig T. MICROSTATELAB: The EEGLAB Toolbox for Resting-State Microstate Analysis. Brain Topogr.; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDelorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchoffelen JM, Gross J. Source connectivity analysis with MEG and EEG. Hum Brain Mapp. 2009;30(6):1857\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan J, Li W, Lin M, Li X, Deng X. Effects of mindfulness and fatigue on emotional processing: an event-related potentials study. Front Behav Neurosci. 2023;17:1175067.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFachner JC, Maidhof C, Grocke D, Nygaard Pedersen I, Trondalen G, Tucek G, et al. Telling me not to worry\u0026hellip; Hyperscanning and Neural Dynamics of Emotion Processing During Guided Imagery and Music. Front Psychol. 2019;10:1561.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MC, Hell D, et al. Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia. Psychiatry Res. 1999;90(3):169\u0026ndash;79.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGreber M, Klein C, Leipold S, Sele S, Jancke L. Heterogeneity of EEG resting-state brain networks in absolute pitch. Int J Psychophysiol. 2020;157:11\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStam CJ, Nolte G, Daffertshofer A. Phase lag index: assessment of functional connectivity from multi channel EEG and MEG with diminished bias from common sources. Hum Brain Mapp. 2007;28(11):1178\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoenig T, Kottlow M, Stein M, Melie-Garcia L. Ragu: a free tool for the analysis of EEG and MEG event-related scalp field data using global randomization statistics. Comput Intell Neurosci. 2011; 2011:938925.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou H. iHBPs-VWDC: variable-length window-based dynamic connectivity approach for identifying hormone-binding proteins. J Biomol Struct Dyn. 2023:1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee JY, Choi CH, Park M, Park S, Choi JS. Enhanced resting-state EEG source functional connectivity within the default mode and reward-salience networks in Internet gaming disorder - CORRIGENDUM. Psychol Med. 2022;52(11):2199\u0026ndash;200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe XQ, Hu JH, Peng XY, Zhao L, Zhou DD, Ma LL, et al. EEG microstate analysis reveals large-scale brain network alterations in depressed adolescents with suicidal ideation. J Affect Disord. 2024;346:57\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMurphy M, Whitton AE, Deccy S, Ironside ML, Rutherford A, Beltzer M, et al. Abnormalities in electroencephalographic microstates are state and trait markers of major depressive disorder. Neuropsychopharmacology. 2020;45(12):2030\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTarailis P, Koenig T, Michel CM, Griskova-Bulanova I. The Functional Aspects of Resting EEG Microstates: A Systematic Review. Brain Topogr. 2024;37(2):181\u0026ndash;217.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNorton DJ, McBain RK, Pizzagalli DA, Cronin-Golomb A, Chen Y. Dysregulation of visual motion inhibition in major depression. Psychiatry Res. 2016;240:214\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCroce P, Zappasodi F, Capotosto P. Offline stimulation of human parietal cortex differently affects resting EEG microstates. Sci Rep. 2018;8(1):1287.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSheline YI, Barch DM, Price JL, Rundle MM, Vaishnavi SN, Snyder AZ, et al. The default mode network and self-referential processes in depression. Proc Natl Acad Sci U S A. 2009;106(6):1942\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLemogne C, le Bastard G, Mayberg H, Volle E, Bergouignan L, Lehericy S, et al. In search of the depressive self: extended medial prefrontal network during self-referential processing in major depression. Soc Cogn Affect Neurosci. 2009;4(3):305\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMayberg HS. Limbic-cortical dysregulation: a proposed model of depression. J Neuropsychiatry Clin Neurosci. 1997;9(3):471\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClancy KJ, Andrzejewski JA, You Y, Rosenberg JT, Ding M, Li W. Transcranial stimulation of alpha oscillations up-regulates the default mode network. Proc Natl Acad Sci U S A 2022; 119(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang Y, Yi Y, Chen Q, Li H, Feng S, Zhou S, et al. Analysis of EEG features and study of automatic classification in first-episode and drug-naive patients with major depressive disorder. BMC Psychiatry. 2023;23(1):832.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaradits M, Bitter I, Czobor P. Multivariate patterns of EEG microstate parameters and their role in the discrimination of patients with schizophrenia from healthy controls. Psychiatry Res. 2020;288:112938.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":false,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-psychiatry","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bpsy","sideBox":"Learn more about [BMC Psychiatry](http://bmcpsychiatry.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bpsy/default.aspx","title":"BMC Psychiatry","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Depressive disorders, Electroencephalogram, Microstates, Functional connectivity, Support vector machine","lastPublishedDoi":"10.21203/rs.3.rs-6170701/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6170701/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Diagnosis of depressive disorders (DD) primarily heavily relies on clinical interviews and subjective symptom reporting, highlighting the urgent need for objective biomarkers. This study aimed to identify novel electroencephalogram (EEG) features, utilizing high-density EEG and source localization techniques, to aid in the diagnosis of DD.\u003c/p\u003e\n\u003cp\u003eMethods: Resting-state EEG data were collected from 115 patients with DD and 43 healthy controls (HCs). Microstate analysis and functional connectivity (FC) analysis were performed to extract EEG features. Statistical analyses were conducted to compare group differences, and a support vector machine (SVM) algorithm was applied to evaluate the diagnostic accuracy of these features.\u003c/p\u003e\n\u003cp\u003eResults: Compared to HCs, DD patients showed a significant increased transition probability from microstate D to B (PFDR = 0.048). Additionally, significant elevations in FC within specific regions of the default mode network (DMN) were observed in the delta-band, theta-band, and beta-band (P \u0026lt; 0.05), as well as between parts of the DMN and the salience network (SN) in the theta-band and alpha-band (P \u0026lt; 0.05). However, none of these FC features survived after FDR correction (PFDR \u0026gt; 0.05). The classification accuracies for distinguishing DD patients from HCs using the SVM classifier were 66.7%, 76.2%, and 81.0% based on microstate features, FC features, and a combination of both, respectively.\u003c/p\u003e\n\u003cp\u003eConclusions: Patients with DD exhibited distinctive microstates and atypical alterations in brain network connectivity. Integrating these features with machine learning algorithms offers a promising approach to improving the objective diagnosis of DD.\u003c/p\u003e\n\u003cp\u003eTrial registration: The study was registered on http://www.chictr.org.cn/ and the registration number was ChiCTR2200057365 (registration date: March 9, 2022).\u003c/p\u003e","manuscriptTitle":"EEG Microstates and Functional Connectivity Abnormalities in Depressive Disorders: Diagnostic Potential and Machine Learning Application","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-09 10:49:41","doi":"10.21203/rs.3.rs-6170701/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-04T14:18:42+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-04T13:33:43+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"244906435551207938515016595893574703394","date":"2026-04-20T13:07:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"14694840478599356651308601799280418444","date":"2026-04-15T18:40:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"154403321374631285425447139183387464415","date":"2025-05-19T14:03:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"89415805706621421429263246461370488586","date":"2025-05-12T14:11:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-05-12T10:32:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"40721087345369688309674198302192911187","date":"2025-05-12T09:55:51+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-02T11:40:12+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-03-18T12:14:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-03-18T04:55:35+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-03-18T04:52:32+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Psychiatry","date":"2025-03-06T12:40:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-psychiatry","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bpsy","sideBox":"Learn more about [BMC Psychiatry](http://bmcpsychiatry.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bpsy/default.aspx","title":"BMC Psychiatry","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e440ce69-62a4-4bb4-a2c8-60120b11cdf1","owner":[],"postedDate":"May 9th, 2025","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-04T14:18:42+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-04T13:33:43+00:00","index":85,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-18T10:38:15+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-09 10:49:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6170701","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6170701","identity":"rs-6170701","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}