Emergence of Individualized Functional Topography in the Neonatal Brain

Emergence of Individualized Functional Topography in the Neonatal Brain | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (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];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Emergence of Individualized Functional Topography in the Neonatal Brain View ORCID Profile Jianlong Zhao , Yuehua Xu , Zaixu Cui , Hongming Li , Lianglong Sun , Xinyuan Liang , Meizhen Han , Zilong Zeng , Qiongling Li , Tengda Zhao , Yong He doi: https://doi.org/10.1101/2025.11.18.689151 Jianlong Zhao 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jianlong Zhao Yuehua Xu 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Zaixu Cui 4 Chinese Institute for Brain Research , Beijing 102206, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Hongming Li 5 Center for Biomedical Image Computing and Analytics, University of Pennsylvania , Philadelphia, PA 19104, USA 6 Department of Radiology, University of Pennsylvania , Philadelphia, PA 19104, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Lianglong Sun 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Xinyuan Liang 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Meizhen Han 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Zilong Zeng 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Qiongling Li 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Tengda Zhao 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: tengdazhao{at}bnu.edu.cn yong.he{at}bnu.edu.cn Yong He 1 State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University , Beijing 100875, China 2 Beijing Key Laboratory of Brain Imaging and Connectomics, Beijing Normal University , Beijing 100875, China 3 IDG/McGovern Institute for Brain Research, Beijing Normal University , Beijing 100875, China 4 Chinese Institute for Brain Research , Beijing 102206, China Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: tengdazhao{at}bnu.edu.cn yong.he{at}bnu.edu.cn Abstract Full Text Info/History Metrics Data/Code Preview PDF Summary The early postnatal period represents a critical phase for establishing the brain’s fundamental architecture. However, how individualized functional topography emerges and shapes long-term cognition remains unknown. Using multimodal neuroimaging data from 419 neonates, we map personalized functional topographies in the first weeks of life and reveal their association with 18-month neurocognitive outcomes. Unique and stable functional topographies are already present at birth, with interindividual variability organized along a conserved sensorimotor-to-association hierarchy. These individualized maps encode early brain maturity and robustly predict cognitive, language, and motor outcomes at 18-month of age, with predictive features concentrated at network boundaries. This functional refinement is structurally anchored by concurrent maturation in cortical myelination and sulcal depth. Preterm birth results in distinctive topographic alterations characterized by accelerated maturation of association networks. Our work establishes a foundation for precision models of early brain development and elucidates its significance for long-term neurocognitive outcomes. Introduction The early postnatal period is a critical and dynamic phase of human brain development, marked by rapid structural and functional reorganization 1 – 3 . During this period, the characteristic spatial architectures, or functional topographies, of nascent functional networks, particularly within the primary visual and somatomotor cortices, are established 4 , 5 . These topographies collectively serve as an essential scaffold for both global neural communication and the protracted maturation of association regions 2 , 4 . Importantly, early functional organization of the brain is not only transient but also lays a foundational blueprint for later cognitive and behavioural capacities 6 , 7 . Alterations to this process, such as those resulting from preterm birth or other perinatal conditions, can disrupt topographic development and increase neurodevelopmental risks 8 – 10 . Thus, the precise mapping of functional topographies in early life is important for understanding typical brain maturation and uncovering the origins of neurodevelopmental disorders. A growing body of evidence highlights the presence of substantial individual variability in neonatal functional organization, particularly within association cortices, spanning from fine-grained voxel-wise patterns to distributed network configurations 11 – 14 . This interindividual variability reflects biologically meaningful distinctions in developmental trajectories 7 , 15 and emerging cognitive abilities 6 , 16 . However, conventional group-level analytic approaches (e.g., group-averaged parcellations) are inherently unable to identify these individual-level features, limiting our ability to precisely track early functional maturation, establish robust brain-cognition relationships, and identify clinically significant alterations in at-risk populations such as preterm infants. As a result, there is an urgent need to characterize personalized functional topographies in the early postnatal brain and elucidate their long-term cognitive and behavioural significance. Recent advances in personalized network modelling have improved our understanding of individual-specific functional architectures in adults 17 – 24 and young people 25 – 27 . Such studies have revealed that individualized functional topographies, when represented as unique probabilistic spatial maps, are highly reliable and exhibit a conserved hierarchical pattern of variability, with the greatest variability in the association cortices and the lowest in sensorimotor cortices 17 , 20 , 25 , 28 . Moreover, these individualized patterns are intricately associated with fundamental cortical properties, including myelination patterns 25 , gene expression profiles 29 , 30 , and evolutionary expansion 25 , and robustly predict individual differences in cognitive performance 21 , 25 , 27 . However, the developmental origins of these personalized functional topographies remain unknown, representing a critical gap in our understanding of how unique functional architectures emerge early in life and shape long-term neurocognitive outcomes. To address this gap, we leveraged the largest multimodal neonatal neuroimaging dataset to date (N = 419) from the Developing Human Connectome Project (dHCP) 31 . Using spatially regularized nonnegative matrix factorization 25 , 27 , 32 , we systematically investigated personalized functional topography during the early postnatal period. The specific aims of this study were to (1) delineate probabilistic functional topographies of neonatal brain networks, (2) quantify their interindividual variability, (3) develop a multivariable framework that uses topographic features to encode early brain maturation and predict 18-month neurocognitive outcomes, (4) elucidate the structural basis of early functional topographic refinements and (5) identify preterm-specific topographic alterations. Together, the findings provide the first comprehensive blueprint of personalized functional topography in the first weeks of life, with fundamental implications for understanding typical and atypical neurodevelopment. Results Data samples Following stringent quality control procedures, we assembled high-quality multimodal neuroimaging, neurocognitive, and behavioural data from 419 term-equivalent infants (including 52 born preterm) sourced from the dHCP 31 . The dataset was stratified into three subsets on the basis of prespecified analytic objectives. (i) Subset 1 (52 term neonates; gestational age [GA] at birth, 37–42 weeks; postmenstrual age [PMA] at scan, 37–44 weeks) data were used to generate initial group-level functional networks, which served as unbiased spatial constraints for subsequent personalized topographic analyses ( Fig. 1a ). (ii) Subset 2 (315 term neonates; GA at birth, 37–42 weeks; PMA at scan, 37–44 weeks) data were used to perform person-specific functional topographic mapping, including the quantification of individual variations, predictive modelling of brain maturity and 18-month neurodevelopmental outcomes, and establishment of the structural basis of early functional development ( Fig. 1a–d ). (iii) Subset 3 (52 preterm infants; GA at birth, 23–36 weeks; PMA at scan, 37–44 weeks) data were used to assess alterations in functional topography in preterm infants ( Fig. 1e ). For comparative neurodevelopmental benchmarking, we included task-free fMRI data from 200 unrelated young adults (22–25 years old) from the Human Connectome Project (HCP) 33 . Demographic characteristics and imaging protocols are detailed in Table 1 . Download figure Open in new tab Fig. 1 Study design and analysis flowchart. a, Workflow for generating group-based and personalized functional topographies of neonatal brain networks. b, Personalized functional topography and interindividual variation. c, Functional topography encodes brain maturation and predicts 18-month neurodevelopmental outcomes. d, Structural basis of functional topography by relating structural features (myelination, thickness, curvature, and sulcal depth) to the contribution weights of functional topography (brain-age and brain-behavioural prediction models) using partial least-square (PLS) analysis. e, Preterm functional topographic alterations in preterm infants were assessed through group comparisons and multivariable classification analyses. View this table: View inline View popup Download powerpoint Table 1 Infant demographics and follow-up neurodevelopmental data Functional topography and individual variation in neonatal brain networks To delineate person-specific functional networks in 315 term-born neonates (Subset 2), we employed NMF 25 , 27 , 32 , a data-driven method that decomposes functional connectivity data into interpretable spatial components. Unlike hard partition approaches that assign each voxel exclusively to a single network, NMF generates probabilistic loadings for each voxel across networks (soft partition), reflecting graded contributions to distinct functional networks. We first established a group-level functional atlas based on NMF using high-quality data from 52 term-born neonates (Subset 1; Fig. 1a ). This group atlas served as an initial reference for individual-level decomposition, ensuring spatial correspondence across participants. Given the known differences between neonatal and adult functional architectures 5 , we avoided the canonical 17-network adult atlas 34 , 35 and instead adopted an 11-network neonatal resolution 10 (validated against alternative 9-/17-network resolutions). Our NMF analysis revealed well-defined neonatal sensorimotor networks ( Fig. 2a ), including a visual network (primary and lateral visual network: Networks 1–2) and a somatomotor network (foot motor network: Network 3; and hand and mouth motor network: Network 4). In contrast, high-order association networks exhibited marked immaturity. The default-mode network fragmented into anterior (DMN-a: Network 5), posterior (DMN-p: Network 6) and lateral (DMN-l and DMN-r: Networks 7-8) components. Frontoparietal and attentional networks appeared as isolated patches (e.g., superior parietal, dorsal frontal, and fronto-limbic: Networks 9-11). These probabilistic functional topographies in neonates align with the findings of prior reports describing well-developed sensorimotor networks and immature association networks during early postnatal development 10 , 36 – 39 ; the present work extends these findings by establishing a probabilistic parcellation framework for neonatal brains. For visualization, we generated a group-based parcellation atlas by assigning each voxel to the network with the highest loading in the probabilistic maps ( Fig. 2b ). Using the group atlas as a spatial constraint, we implemented individualized NMF to derive eleven person-specific functional networks for 315 neonates in Subset 2 (Subset 1 was excluded to avoid data leakage). To validate the efficacy of individualized network parcellations, we assessed the functional homogeneity of each network, quantified as the average Pearson’s correlation coefficients of time series between all pairs of voxels within a given network. Compared with the group-level atlas, individualized networks exhibited greater functional homogeneity (t = 20.70, P = 2.33 × 10□□, Cohen’s d = 1.98; Fig. 2c ). Download figure Open in new tab Fig. 2 Functional topographies of neonatal brain networks. a , Group-level probabilistic network topographies. Voxel-wise loadings represent the degree of network membership, with warmer colours indicating higher loadings. b, Hard parcellation obtained by assigning each voxel to the network with the highest loading. c, Compared with the group-level atlas, the individualized network maps exhibited significantly higher within-network functional homogeneity. ***, P < 0.001. We next investigated interindividual variability in neonatal functional topography. Visual inspection revealed that individual variations were apparent in the association cortices, such as those in the dorsal frontal network (Network 10), and that the primary networks, such as the foot motor network (Network 3), were broadly conserved across neonates ( Fig. 3a ). To quantify these interindividual variabilities in functional topography, we computed the median absolute deviation (a nonparametric statistic of variance) of probabilistic loading across neonates for each network 25 , 40 . We found that the association cortices exhibited greater interindividual topographic variability than the primary cortices ( P spin = 0.017) ( Fig. 3b ). Further analyses of four topographical properties—position, size, overlap, and homogeneity 21 , 41 —also confirmed that the association networks were more variable across neonates than the primary networks (position: P spin = 0.002; size: P spin = 0.026; overlap: P spin < 0.001; homogeneity: P spin < 0.001; Fig. 3b ). Notably, this neonatal variability pattern was spatially correlated with adult interindividual differences (HCP dataset, r = 0.46, P spin < 0.001; Fig. 3c ), suggesting a conserved organizational hierarchy. Importantly, our findings trace the origins of these topographical variations, offering the earliest evidence for their ontogenetic establishment. Finally, to delineate fine-grained variability patterns along the cortical sheet, we performed boundary concentration analysis by assessing the spatial correlations between individual variability maps and network loading maps. We observed significant negative correlations at the global (r = -0.61, P spin < 0.001) and network levels (all P spin < 0.05, except Network 10) ( Fig. 3d ), indicating that the greatest interindividual variations tended to be located at network boundaries where the network loadings were smaller. This effect was more pronounced in the primary networks than in the association networks ( P spin = 0.031; Fig. 3d ). Download figure Open in new tab Fig. 3 Interindividual variability in neonatal functional topography. a, Visual inspection of individualized voxel-wise probabilistic network maps across representative neonates. b, Quantification of interindividual variability at the global, network, and topographic levels. c, Spatial correlation of functional topographic maps between neonatal and adult brains. d, Boundary concentration analysis by correlating interindividual topographic variability and network loadings of functional topography at both the global and network levels. Functional topography undergoes refinement in the early postnatal period and encodes brain maturity Having identified the interindividual topographic variability in neonatal functional networks, we next investigated how person-specific functional topography encodes brain maturation in 315 neonates (Subset 2). First, we performed mass univariable analyses to characterize developmental changes in both total and voxel-wise network organization. (i) For the total network representation, we summed voxel-wise loadings for each probabilistic network and modelled age-related changes using generalized additive models (GAMs), controlling for sex, mean framewise displacement (FD), and scan-birth age intervals. The lateral visual network (Network 2) exhibited a nonlinear increase in total representation with age ( Fig. 4a , P = 0.004, FDR corrected), whereas the fronto-limbic network (Network 11) gradually decreased with age ( Fig. 4a , P = 0.02, FDR corrected). (ii) Voxel-wise GAM analyses for each probabilistic network revealed significant age-dependent refinement in network loadings in the visual (Networks 1–2), somatomotor (Network 4), superior parietal (Network 9), and frontal-limbic (Network 11) networks (Gaussian random field (GRF) correction at voxel-level P < 0.001, cluster-level P < 0.05) ( Fig. 4b ). No significant sex effect was observed for either global or voxel-wise functional topography. Download figure Open in new tab Fig. 4 Functional topography undergoes progressive refinement and encodes brain maturity during the early postnatal period. a, Developmental trajectories of total network representation across scan ages. The total representation was quantified as the sum of the voxel-wise probabilistic loadings for each network. b, Developmental trajectories of the voxel-wise functional topography. Analyses were conducted at the voxel level ( P < 0.001) and corrected at the cluster level ( P < 0.05) using Gaussian random field theory. c, Functional topography encodes brain maturation using Ti-PCA and SVR frameworks. The bars indicate network-level weights (red: primary; yellow: association). Dashed outlines denote networks below the 50% contribution threshold. d, Network- and voxel-level contribution weights derived from the brain-age prediction model. Primary and association networks are colour coded. Dashed outlines denote networks below the 50% contribution threshold. e, Correlations between contribution weights and network loading across voxels. f, Correlation between contribution weight and interindividual topographic variability. Dashed outlines denote networks below the 50% contribution threshold. Second, to capture the high-dimensional reconfiguration of the functional topography, we developed a multivariable prediction framework that combines topography-independent principal component analysis (Ti-PCA) with support vector regression (SVR). Briefly, PCA was performed separately within each network in the training set while preserving feature independence across networks. The resulting PCA components were integrated into an SVR model as feature sets to predict postmenstrual age at scan. Using 10-fold cross-validation, we evaluated prediction accuracy via the Pearson’s correlation coefficient between predicted age (“brain maturity index”) and chronological age in the testing set after regressing out the effects of sex, mean FD, and scan-birth age intervals. This model achieved significant prediction accuracy (r = 0.46, permutation test, P perm < 0.001), and its robustness was confirmed across 100 repeated random folding analyses (mean r = 0.51, P perm < 0.001) ( Fig. 4c ). Although single-network features yielded significant prediction accuracies (r = [0.15–0.40], P perm < 0.002), their inferior performance relative to that of TiPCA underscores the critical advantage of multinetwork integration for modelling brain maturity. High-contributing networks (those with top 50% predictive accuracy) included all primary sensorimotor networks (Networks 1–4) and a subset of association networks (e.g., DMN-p, Network 6, and the fronto-limbic network, Network 11) ( Fig. 4d ). Voxel-wise feature weights were negatively correlated with network loading (r = -0.38, P spin < 0.001; Fig. 4e ) and positively correlated with interindividual variability (r = 0.46, P spin < 0.001; Fig. 4f ). Taken together, these results suggest that early postnatal development is primarily encoded at the network boundaries of the sensorimotor system and a few association systems and is further shaped by person-specific network variability. Neonatal functional topography predicts neurodevelopmental outcomes at 18 months We next assessed whether personalized functional network topography at birth could predict individual neurodevelopmental outcomes at 18 months. In Subset 2, neonates (n = 249) participated in a follow-up visit at 18 months, where cognitive, language, and motor abilities were evaluated using the Bayley Scales of Infant and Toddler Development, 3rd Edition (Bayley-III) 42 . Multivariable analysis using Ti-PCA and SVR revealed that the individual functional topography significantly predicted individual cognitive (r = 0.39, P perm < 0.001), language (r = 0.29, P perm < 0.001), and motor (r = 0.37, P perm < 0.001) scores (10-fold cross-validation; covariates: sex, mean FD, scan age, and scan-birth age intervals) ( Fig. 5a ). This prediction remained robust following 100 random-fold validations (cognition: mean r = 0.34; language: mean r = 0.27; motor: mean r = 0.34; all P perm < 0.001; Fig. 5a ) and outperformed single-network predictions (cognition: r range = [0.10, 0.32]; language: r range = [0.11, 0.27]; motor: r range = [0.13, 0.31]). High-contributing networks across the three prediction models included the visual (Network 1-2), DMN-a (Network 5), and superior parietal (Network 9) networks, while domain-specific involvement was observed with the DMN-p (Network 6) and somatomotor (Network 3-4) networks ( Fig. 5b ). Voxel-wise contributing weights were negatively correlated with network loading (cognition: r = -0.39; language: r = -0.40; motor: r = - 0.46; all P spin < 0.001) and positively correlated with individual variability (cognition: r = 0.48; language: r = 0.47; motor: r = 0.52; all P spin 0.05; covariates: sex, mean FD, scan age, and scan-birth age interval). Taken together, these results suggest that the functional topography-based predictions for neurodevelopmental outcomes were driven mainly by network boundaries with high across-neonate variability. Download figure Open in new tab Fig. 5 Neonatal functional topography predicts neurodevelopmental outcomes at 18 months. a, Multivariable prediction of neurodevelopmental outcomes at 18 months using personalized neonatal functional topography. Shaded areas represent 95% confidence intervals. The bottom panels display the prediction accuracy distributions across 100 repeated random cross-validation iterations. b, Contribution weight maps for the prediction of cognitive, language, and motor outcomes. The bars indicate network-level weights (red: primary; yellow: association). Dashed outlines denote networks below the 50% contribution threshold. Brain maps depict voxel-wise predictive weights. For neonates (n = 245) in Subset 2, the Quantitative Checklist for Autism in Toddlers (Q-CHAT) was completed at the 18-month follow-up visit to assess atypical behavioural risk. Multivariable analysis using Ti-PCA and SVR revealed that person-specific functional topography predicted individual Q-CHAT scores (10-fold cross-validation: r = 0.29, P perm < 0.001; 100 random-fold validation: mean r = 0.30, P perm < 0.001). High-contributing networks included the primary sensorimotor (Network 2-4) and association (Networks 5, 7 and 9) networks. However, univariable GAM analysis revealed no significant correlation between the total loading of each network and the 18-month Q-CHAT score after FDR correction. Functional topographic refinement in the neonatal brain is anatomically constrained by structural maturation We sought to delineate whether neonatal functional topographic refinement during the early postnatal period is anatomically constrained by cortical structural properties. Using structural images from 301 term-born neonates (Subset 2), we first generated individual cortical maps of four anatomical features, namely, cortical myelination, cortical thickness, surface curvature, and sulcal depth. We subsequently used GAMs to examine the spatial maturation of these structural features during the early postnatal period, with scan age as the main factor and sex, mean FD, and scan-birth age interval as covariates ( Fig 6a ). Cortical myelination exhibited pronounced age-dependent changes, particularly in the sensorimotor and insular cortex (P < 0.05, FDR corrected). Cortical thickness, curvature, and sulcal depth exhibited age-related changes in the boundaries of the medial foot motor cortex and anterior DMN (P < 0.05, FDR corrected). Download figure Open in new tab Fig. 6 Functional topographic refinement in the neonatal brain is structurally constrained by cortical anatomical maturation. a, Cortical structural maturation pattern during the early postnatal period. F statistics for the effects of scan age are shown for myelination, sulcal depth, cortical thickness, and curvature. b, Associations between structural maturation and functionally defined, network-level contribution weights from the brain-age prediction models. c, Associations between structural maturation and contribution weights from the brain–behaviour prediction models. In panels (b) and (c), violin plots display whole-brain and network-wide structure□function associations. Asterisks indicate networks with significant structure–function associations after Bonferroni correction. Red and yellow violins represent primary and association networks, respectively. Scatter plots show structure–function correlations at the network level, and bar plots depict structural loadings for sulcal depth, myelination, thickness, and curvature. Next, we performed partial least-square (PLS) analyses to assess the latent relationship between these maturation maps ( F values of the age effect) and the contribution weights of network-specific functional topography in the brain-age and brain-behaviour prediction models. To mitigate potential biases due to the unequal network size, we applied bootstrap resampling (100 iterations) to ensure the same number of voxels across networks prior to the PLS analysis. The contributing feature weights of functional networks in brain age were correlated with age-related structural maturation (foot motor: r = 0.37, P perm = 0.001; DMN-p: r = 0.39, P perm = 0.001; Bonferroni corrected), with the greatest loading value at the sulcal depth and in the myelin maps ( Fig 6b ). The contributing feature weights of functional networks in brain-behavioural predictions were correlated with structural maturation (DMN-a: r = 0.43, P perm = 0.001; foot motor: r = 0.45, P perm = 0.001; Bonferroni corrected), with myelination and sulcal depth emerging as the strongest loading factors ( Fig. 6b ). These results suggest that the refinements of functional topography in the neonatal brain are structurally supported by cortical maturation, particularly by myelination and sulcal depth. Altered network topography and accelerated functional maturation in preterm infants Previous studies have demonstrated that compared with term-born controls, preterm infants exhibit altered patterns of functional brain maturation 10 , 43 , 44 . Here, we systematically investigated functional topographic alterations among 52 preterm infants scanned at term-equivalent age (Subset 3). For comparison, we used the neuroimaging data of 73 term-born neonates (Subset 2), matched for scan age, sex, and mean FD. Preterm infants had lower birth weights than term-born controls did (Z = 8.7; P < 0.001, Cohen’s d = 2.64) but had comparable 18-month neurodevelopmental outcomes (Table S1). The Ti-PCA-based SVM analysis robustly distinguished preterm infants from term infants, achieving 82% accuracy (sensitivity: 75%; specificity: 86%; 1000 permutation tests: P perm < 0.001; 100 random fold cross validations: mean accuracy = 0.79, P perm < 0.001) after controlling for sex, mean FD, and scan age ( Fig. 7a ). The most discriminative networks were primarily in several association systems, including the fronto-limbic network (Network 11), superior parietal network (Network 9), and DMN (Networks 5–8) ( Fig. 7a ). These voxel-wise contribution weights were negatively correlated with network loading (r = -0.33, P spin < 0.001) and positively correlated with individual variability (r = 0.48, P spin < 0.001). Mass univariable GAM analyses (covariates: sex, mean FD, and scan age) also confirmed atypical functional topography in these association networks in preterm infants compared with term-born controls. Download figure Open in new tab Fig. 7 Altered network topography and accelerated functional maturation in preterm infants. a, Multivariable classification between term and preterm infants on the basis of individualized functional topography. Classification performance was evaluated using 10-fold cross-validation and repeated across 100 random folds. Network- and voxel-level contribution weights are shown. b, Brain age gap analysis between term and preterm infants. The BAG was computed at the whole-brain and network levels. Right, association between the BAG and motor scores at 18 months. c, Structural differences between term and preterm infants across four cortical features: myelination, sulcal depth, cortical thickness, and curvature. PLS analysis was used to assess the association between structural differences and functional contribution weights. Structural loadings are shown for each feature. We further assessed whether preterm infants exhibited delayed or accelerated functional maturation ( Fig. 7b ). Using the TiPCA model trained on term-born neonates, we obtained significant predictions for preterm infants (r = 0.41, P = 0.002 for the whole brain; r = [0.31, 0.39], P <= 0.02 for specific networks). After the personalized brain age gap (BAG) (predicted minus actual age) was computed with rigorous bias correction 45 , 46 , preterm infants exhibited greater BAGs than term-born infants did (whole-brain: t = 2.41, P = 0.016, Cohen’s d = 0.36; network-level: DMN-p: t = 3.72, P = 0.0002, Cohen’s d = 0.56; frontal-limbic: t = 3.45, P = 0.0006, Cohen’s d = 0.52) ( Fig. 7b ). These findings indicate accelerated topographic maturation in preterm infants during early postnatal development. The DMN-p BAG was positively correlated with 18-month motor scores in preterm infants (r = 0.54, P = 0.001; covariates: sex, mean FD, scan age, and the scan-birth interval) ( Fig. 7b ), but the whole-brain BAG and other network BAG were not significantly correlated. Finally, structural analyses using univariable GAMs revealed significantly lower myelination in the orbital and medial frontal and temporal parietal cortex in preterm infants than in term-born controls ( P < 0.05, FDR corrected) ( Fig. 7c ). Cortical thickness, curvature, and sulcal depth exhibited age-related changes in the boundaries of the dorsal frontal and temporal cortex (P < 0.05, FDR corrected). Multivariable PLS analysis was used to test whether structural differences ( Z values) were correlated with the functionally defined feature contributions of the classification model. We showed that the DMN-l contribution weights were correlated with structural abnormalities (DMN-l: r = 0.34, P perm = 0.002, Bonferroni corrected), which were primarily driven by the cortical development of myelination and sulcal depth ( Fig. 7c ). Sensitivity analyses Our results were validated using multiple analytical strategies, including assessments of sample size, scan duration, network resolution, and head motion control. (i) To evaluate the effects of head motion, the analysis was repeated using neonatal data under a stricter FD threshold (mean FD < 0.3 mm). (ii) To examine the influence of scan duration, neonatal functional topographies were reconstructed using truncated time series ranging from 1 to 15 minutes (across seven intervals) and compared against full-length data. (iii) To assess the impact of sample size on brain age and brain behaviour prediction models 47 , 48 , bootstrapped subsampling was performed across logarithmically spaced sample sizes (n = 25–284 neonates for brain age; n = 25–224 neonates for behaviour), with prediction analyses repeated 100 times per sample size. (iv) To test the reproducibility of the results, a split-half validation approach was employed, in which subsets were matched by neonatal scan age. (v) To determine the effect of sample selection on initial group-level atlases, we randomly selected subsets of term-born neonates (ranging from 10 to 300 scans). (vi) To evaluate sensitivity to network resolution, neonatal functional topographies were regenerated using parcellation schemes with k = 9 37 , 49 and k = 17 34 , 50 , 51 . Across all analyses, our findings remained highly consistent and robust, demonstrating minimal influence from variations in head motion thresholds, scan duration, sample size, and network resolution. These results confirm the reliability and stability of our conclusions across diverse analytical conditions. Discussion Using personalized network modelling, we mapped developing cortical functional topography during the initial postnatal weeks. We found that while primary functional networks are already well established at birth, high-order association networks are present only as early prototypes. We observed that the refinement of these nascent networks occurs at their boundaries in an individualized manner and that this refinement process collectively predicts neonatal topography for both preterm and term-born infants, revealing an accelerated maturational pattern in the association networks of preterm infants. These patterns of functional maturation are significantly correlated with indicators of concurrent structural development, particularly in terms of cortical myelination and sulcal depth. Together, our findings provide mechanistic insights into how the initial functional configuration of the neonatal brain provides a scaffold for the development of long-term cognitive ability. The neonatal human brain already has complex neural configurations. At the cellular level, processes such as neurogenesis, synaptogenesis, and axon growth establish a foundational blueprint, priming the brain for postnatal axonal pruning and neural consolidation 1 , 52 . Functionally, persistent neuronal activity refines early neural circuits into organized spatial layouts (functional topography) 36 , 38 , 39 and reflects high postnatal plasticity 53 . Building on this, we identified largely adult-like functional networks in the primary sensorimotor cortex, alongside still-maturing networks in the association cortex. Specifically, primary visual, somatomotor, and supplementary motor networks closely resembled their mature counterparts. In association networks, we delineated three fine-grained frontal networks 10 , 37 and two distinct DMN components, moving beyond prior coarse descriptions 10 . Notably, the emergence of high-order networks in neonates remains debated, with some studies reporting difficulty in detection 37 , 54 and others confirming their presence at birth 36 , 51 , 55 . Sylvester and colleague emphasized the importance of preserving long-distance connections to identify advanced functional networks 51 , whereas others noted neonatal adult differences in cross-cortical connectivity 56 . Our study demonstrates that effective machine learning algorithms such as NMF can recognize these networks at the individual level. Notably, these high-order networks, even in primordial forms, significantly predict 18-month neurodevelopmental outcomes at 18 months. For instance, the anterior DMN and lateral parietal network contribute to cognitive, language, and motor scores, underscoring their functional relevance from early development. We demonstrate that the fine-grained individual variation in neonatal function topography differs fundamentally between association networks and primary networks. While primary networks vary predominantly at their boundaries, association networks exhibit heterogeneity in both network cores and boundaries. This divergence likely stems from distinct cortical organizational principles and maturation dynamics 10 , 37 . Unlike the primary cortex, association networks are architecturally optimized for higher-order cognition through complex synaptic wiring, enhanced plasticity, and elevated metabolic demands 57 , 58 , potentially supported by fine-scale functional subnetworks with distributed representations 59 . Developmentally, association networks remain in a dynamic state of spatial refinement during the neonatal period, shaped by genetic and environmental factors 1 , 60 , with substantial individual variation 12 – 14 . In support of this, Xu et al . 13 reported that functional network variability decreases with age differentially between primary and association systems. Given the prolonged maturation of the association cortex into adolescence 58 , future longitudinal studies integrating genetic and neuroimaging data across extended developmental windows will be crucial to elucidate these mechanisms. The emerging “brain age” concept has proven valuable for characterizing typical and atypical brain maturation by quantifying the discrepancy between neuroimaging-predicted age and chronological age 61 , 62 . Previous studies have demonstrated that cortical functional connections, particularly in the sensorimotor cortex, can accurately predict an individual’s scan age 15 , 63 . Our findings further establish individualized functional topography as a key marker of neonatal brain maturation. Notably, we found that spatial refinement at network boundaries, especially in primary networks, contributes most substantially to brain age prediction, aligning with evidence that primary cortices undergo more pronounced early development than high-order regions do 38 , 64 . Histological studies have confirmed that primary cortices experience accelerated neurogenesis, synaptogenesis and axon growth around birth 65 , supporting the emergence of core visual and motor functions 52 . This early microstructural maturation may establish stable functional topography at network cores while maintaining developmental sensitivity at boundaries, creating a characteristic spatial maturation gradient. Understanding the neurodevelopmental origins of diverse behavioural outcomes is critically important 1 , 66 . Our results demonstrate that individualized functional topography measured at birth accurately predicts cognitive, language, and motor abilities at 18 months of age. Several factors may explain this predictive relationship. First, despite its relatively coarse anatomical scaffold, the neonatal brain has already undergone substantial development 52 , including neurogenesis, synaptogenesis, axonal growth, and initial myelination. These processes collectively establish the foundation for subsequent cognitive capabilities 67 , 68 . Second, although immature compared with adult benchmarks, neonatal functional configurations exhibit sufficient individual specificity to support identification 69 , 70 and predict later learning 69 and cognition outcomes 71 . Third, the early postnatal period represents a window of heightened plasticity during which brain function rapidly adapts to environmental inputs 72 , enabling remarkable learning abilities even within hours of birth 53 . Notably, the most consistent predictive voxels for various neurodevelopmental outcomes clustered around network boundaries, particularly in primary sensorimotor networks, suggesting that these regions may function as common computational cores supporting diverse types of behavioural development. Previous research has indicated that compared with core regions, border zones of primary systems display distinct connectivity patterns, morphological features, and cytoarchitectonic organization 73 . Functionally, these border areas predominantly occupy unimodal-to-limbic hierarchical positions 73 , which are crucial for stabilizing and integrating inputs along the sensorimotor-to-association gradient 73 . In our data, supplemental motor area boundaries showed prominent refinement—a region recently identified as part of an “action-mode network” implicated in action planning 74 . Our findings align with established developmental sequences: infants initially develop visual and motor abilities to support facial recognition, vocalization, and head control 75 , which are abilities that foster secure attachment and facilitate higher-order skill acquisition 76 . Moreover, association networks, including the DMN and lateral parietal cortex, differentially predicted specific 18-month outcomes, which is consistent with their known involvement in social cognition 77 , rapid learning 53 , and emotional processing 78 . The mechanisms through which these networks support behavioural development are undoubtedly complex; future studies should integrate network communication models with task-based fMRI to elucidate how distributed neural interactions give rise to individual differences 79 . Overall, the first two years of life represent a critical period for establishing lifelong behavioural foundations. Our findings provide potential neuroimaging biomarkers for tracking brain–behaviour relationships during this crucial window and may help elucidate the neural mechanisms underlying developmental neuropsychiatric disorders. Preterm birth has been consistently linked to widespread neural alterations, including reduced cortical volume, microstructural changes, and disrupted white matter organization at term-equivalent age 80 , 81 . Functional abnormalities have also been reported in connectivity, network topology, and dynamics in preterm infants 10 , 82 . Extending these findings, we demonstrate that individualized functional topography reliably distinguishes preterm infants, with association networks contributing most significantly. These results align with dHCP-based reports of functional alterations in frontal parietal networks 44 . Early exposure to extrauterine stimuli may place exceptional adaptive demands on developing association systems 83 , which could bear the primary burden of accommodating this abrupt developmental transition. Interestingly, although prior studies describe marked abnormalities in primary visual or motor systems 10 , 43 , our approach revealed limited effects in these regions, possibly because the core spatial components of primary networks are already established and remain stable. We also reported accelerated functional maturation in preterm infants, which is consistent with reports of white matter 84 , 85 and cortical development 86 . Potential mechanisms include birth as an environmental trigger altering synaptic trajectories 87 , early visual experience accelerating parvalbumin expression 88 and enhanced early nutrition. Our model enables the individualized quantification of this accelerated maturation, offering a potential neuroimaging indicator of prematurity-related developmental pathways. Several potential limitations of this study should be acknowledged. First, the analysis was restricted to cortical regions. Given that subcortical areas and the cerebellum also undergo rapid structural and functional development during the neonatal period 1 , future studies could incorporate these regions to examine how individualized functional topography evolves across the whole brain and its association with long-term neurocognitive outcomes. Second, all the neonatal scans in the dHCP dataset were performed during natural sleep. Variations in sleep state may introduce bias in the estimation of functional connectivity and individual variability 89 ; however, acquiring fMRI data from awake neonates remains highly challenging 90 . Third, previous research suggests that 18 months of age represents an early stage at which initial symptoms of certain neurodevelopmental disorders, such as autism, may develop 91 . This underscores the importance of collecting longitudinal data from atypically developing populations to investigate how neonatal functional architecture influences the subsequent onset of neuropsychiatric conditions. Materials and methods Participants and data acquisition The whole dataset comprises a high spatial–temporal resolution of 367 term-born neonates and 52 preterm infants as part of the Developing Human Connectome Project 31 . All infants were scanned during natural sleep without sedation to obtain structural and resting-state fMRI data using a 3T Philips Achieva scanner. Bayley-III scores were assessed in 249 infants at 18 months of age to capture individual developmental outcomes in motor, language, and cognitive domains 42 . The Q-CHAT score for 245 infants at 18 months of age was also assessed to measure atypical social, sensory, and repetitive behaviour risks 92 . Written consent was obtained from the guardian of each baby. The dataset was further stratified into three groups. Subset 1 included 52 term-born neonates (36 females; GA at birth, 37–42 weeks; PMA at scan, 37–44 weeks, 15 minutes scan) and was used to construct a typical neonatal group atlas; Subset 2 included the remaining 315 term-born neonates (134 females; GA at birth, 37–42 weeks; PMA at scan, 37–44 weeks, 15 minutes scan) and was used to reconstruct individualized functional topography to delineate functional variability and to conduct prediction analysis on brain maturity and neurocognitive outcomes; and Subset 3 included 52 preterm infants (19 females; GA at birth, 23–36 weeks; PMA at scan, 37–44 weeks, 15 minutes scan) and was used to conduct classification analysis between preterm and term infants. Details are provided in Table 1 . For comparison purposes, we also included 200 adults from the public HCP S1200 dataset 33 (73 females, 22–25 years old, 14 minutes scan). We selected the brain scans of the 200 youngest healthy adults (73 females; ages 22–25 years; 14-minute scans) from all the participants. Written informed consent was obtained from all the subjects, and the scanning protocol was approved by the Institutional Review Board of Washington University in St. Louis, MO, USA (IRB #20120436). Details are provided in Supplementary Section 1. MRI data preprocessing For all the fMRI images in the dHCP dataset, we performed three preprocessing stages: minimal preprocessing, customized spatial normalization, and final-stage preprocessing. The minimal preprocessing steps were performed by the dHCP organizer and have been validated in prior studies 93 . To obtain good spatial correspondence, spatial normalization steps were designed by generating a customized template with a Serag template 94 at 37 weeks initially using all the brain scans from Subset 1. Individual T2w images and tissue masks were nonlinearly registered to the customized template. Final-stage preprocessing was performed using GRETNA v2.0.0, and included smoothing, linear detrending, nuisance variable regression, scrubbing, and temporal filtering. Notably, Friston’s 24 head motion parameters and average signals were removed using multivariable linear regression, and spike regression-based scrubbing (0.5 mm displacement) 95 , 96 was performed to control for the effect of head motion. For the HCP dataset, all functional images initially underwent minimal preprocessing using the HCP public pipeline 97 . Following this, similar processing steps as those for the dHCP dataset were performed. Generating individualized brain networks 1) Group network identification To make a typical representation of the broadly consistent functional network across the neonatal stage, we first generated a group network atlas, which was then used as initialization for subsequent individualized network generation. As previously described 25 , 32 , we used NMF to derive these group networks. The NMF approach employs a group consensus regularization term that preserves the interindividual correspondence of labels and locality between functional networks and ensures that the decomposition is robust to imaging noise (see 32 for details). We set the number of networks as 11 5 , 10 in the main analyses and validated it at 9 and 17. To reduce the influence of outliers, a bootstrap strategy was used to perform group-level decomposition on 80% of the subjects randomly selected each time, which was repeated 50 times. During each iteration, a time series matrix with 92 000 rows (time points) and 6905 columns (voxels) was constructed, and a network loading matrix was constructed, which yielded 50 different group atlases in total. We combined these atlases into a robust loading matrix V using spectral clustering 32 . The resulting group-level network loading matrix V had N rows (network number) and 6905 columns (voxel number), representing the loading of each given voxel. 2) Individualized network mapping We mapped each individualized specific network atlas using a second-time regularized NMF based on the identified group networks as initialization and individual fMRI time series (2300 × 6905) as input, which yielded a loading matrix V (11 × 6905 matrix) for each participant. This probabilistic (soft) definition was converted into discrete (hard) network definitions for display by labelling each voxel according to its highest loading. Individual variation in neonatal functional network topography To represent the individual variation in functional network topography, we first computed the voxel-wise variability by calculating the median absolute deviation of the probability network loading across neonates for each voxel and summing across networks to generate a whole-brain variability map. Group differences between primary and association networks were assessed using spin-based permutation testing. Next, four network-level topographic properties were evaluated: 1) position variability—we calculated the average Euclidean distance among the parcel centres (centre of mass) across subjects; 2) size variability—we calculated the standard deviation of size (the number of voxels of a region) across subjects; 3) overlapping variability—we calculated the mean Dice coefficient deviation between each pair of subjects; and 4) homogeneity variability—we calculated the standard deviation of homogeneity across subjects (for a detailed algorithmic workflow, please refer to 41 ). All comparisons between primary and association networks were tested using 10,000 spin permutations. Spatial permutation testing (spin test) To assess whether the individual variation in functional topography is significantly constrained by the primary-association system, we conducted a permutation test correcting for spatial autocorrelations (10,000 times) 98 . In each permutation, we randomly shuffled the across-individual variance Z-map and created a spatially autocorrelated surrogate Z-map by reintroducing the spatial autocorrelation characteristic of the original, nonpermuted data. This map was divided into association system and primary system variations, which were averaged, and the deviation between them was calculated. After 10,000 iterations, we generated a null model and estimated the P value by comparing the real deviations with those in the null model. Associations of network topography with development, neurodevelopmental outcomes at 18 months, and preterm birth Given the nonlinear nature of brain development, we used GAMs to capture the associations between network topography and scan age while controlling for sex, mean FD, and scan-birth age intervals. To evaluate 18-month neurodevelopmental outcome associations, GAMs were similarly fitted with outcome scores as dependent variables, adjusting for sex, mean FD, scan age, and the time interval between birth and the scan. To evaluate preterm birth associations, preterm status (preterm vs. term) was included as the primary group factor, with sex, mean FD, and scan age as covariates. These models were fitted at both the network and voxel levels. At the network level, FDR correction was employed to control for family-wise error. At the voxel level, the significance threshold was set at P < 0.001 at the voxel level, with Gaussian random field (GRF) correction at the cluster level of p < 0.05 99 . Multivariable topographic network-level prediction framework (TiPCA model) To assess whether and how individualized functional network topography predicts brain maturity and neurodevelopmental outcomes and to classify preterm infants from their term-born counterparts, we developed a multivariable TiPCA prediction model approach based on the SVR framework. This approach involves performing feature reduction by separately performing PCA within each network before model training. We validated the robustness of the prediction model by testing prediction accuracy using repeated random 10F-CV, repeating 10F-CV 100 times and averaging the correlation to determine overall prediction accuracy. To identify high contributing networks, we separately assessed the prediction power of each network and identified high contributing networks as those whose prediction accuracy ranked in the top 50% across all the networks. To further depict the spatial topography of high-contributing voxels, we calculated the voxel-wise feature contribution weights within the high-contributing network and summed the absolute weights across networks. The TiPCA framework was consistently used for the prediction of individual scan age, Bayley-III scores and Q-CHAT scores with a 10-fold cross-validation strategy. Different confounding variables were considered for each prediction. To predict brain maturity, we regressed out the confounding variables of sex, mean FD, and the time interval between birth and the scan from both functional topography and scan ages. To predict behavioural assessments, we regressed out confounding variables (age at scan, sex, mean FD, and scan-birth age intervals) from both functional topography and neurodevelopmental outcomes. Cortical structural maturation and structure□function association analysis To investigate whether neonatal functional topographic refinement is constrained by cortical structural maturation, we employed structural MRI scans from 301 term-born neonates (Subset 2). For each participant, we derived individual cortical surface maps for four anatomical features: cortical myelination (T1w/T2w ratio), sulcal depth, cortical thickness and surface curvature. We assessed age-related changes in each structural feature using vertex-wise GAMs, with scan age as the main factor and sex, mean FD, and the interval between birth and scan age as covariates. The resulting F-statistic maps quantified the spatial pattern of structural maturation across the cortex. Significance was determined using vertex-wise FDR correction at q < 0.05. To examine structure function relationships, we performed PLS analyses between the structural maturation maps (F values) and the contribution weights of functional topography derived from predictive models of brain age and behavioural outcomes. Specifically, we computed spatial correlations between functional feature weights (from the TiPCA model) and the structural maturation maps in the PLS model at both the global and network levels. We summarized the PLS loadings across the four structural features to identify the dominant contributors. Bonferroni correction was applied across networks. Identifying preterm infants and estimating personalized brain age gaps To classify term-born and preterm infants at term-equivalent age, we employed a similar TiPCA classification approach with a 10-fold cross-validation strategy using Subset 2 and Subset 3 (the 52 preterm infants and matched 73 term neonates (scan age, sex, and head motion matched)). Before classification, we regressed out confounding variables (age at scan, sex, and mean FD) from functional topography. We then evaluated the network-wise feature weights for preterm and term classification to determine their contributions to the model. To quantitatively assess the degree of abnormal maturation in preterm infants, we used preexisting brain maturity prediction models trained on term-born neonates to predict the PMA at scan for each preterm-born infant as the brain age. By calculating the difference between the predicted postmenstrual age and the actual postmenstrual age, we derived the personalized brain age gap (BAG) for each preterm infant. To adjust for prediction biases 45 , we regressed out chronological age from all estimated BAGs and used residuals to define adjusted age gaps. Regression coefficients from the term-born cohort were applied to adjust the BAGs in the preterm group. To determine whether brain maturation is accelerated in preterm infants relative to term-born infants, we applied a two-sample t test to compare the BAG between the two groups. We also examined associations between the BAG and 18-month neurodevelopmental outcomes using partial correlations while controlling for sex, scan age, FD, and the time interval between birth and the scan. Finally, we explored whether alterations in functional topography in preterm infants were constrained by structural abnormalities. Individual cortical maps of myelination, thickness, curvature, and sulcal depth were derived and aligned to a common surface template. Group differences were assessed using GAMs, with sex, mean FD, and scan age as covariates. Statistical significance was determined using FDR correction (q < 0.05). A similar PLS analysis was used to test the association between group-level structural abnormalities (Z statistics of group differences) and network-wise contribution weights from the classification model. Sensitivity analysis We conducted multiple sensitivity analyses to ensure the robustness of our findings. Specifically, we (i) repeated all analyses using a stricter head motion threshold (mean FD < 0.3 mm); (ii) reconstructed group and individualized functional topographies using truncated time series of varying scan durations (1–15 min); (iii) evaluated the effect of sample size on brain age and brain–behaviour predictions through bootstrapped subsampling across logarithmically spaced sample sizes; (iv) performed split-half validation by dividing neonates into two age-matched subsets; (v) regenerated group-level atlases using randomly selected subsets of term-born neonates (n = 10–300); and (vi) tested sensitivity to network resolution using parcellations with 9 and 17 networks. Further methodological details are provided in Supplementary Information, Section 6. Data availability All data required for reproducing our findings have been publicly available, including the group-level network, the individualized functional networks, individual variability, brain maturity prediction model, neurodevelopmental outcomes prediction model, structure–function coupling, Preterm functional topographic alterations, and the data for visualizing main figures. They are stored in a publicly accessible cloud repository (github: https://github.com/zhaohuaxishi1/neonatal_Individual_Parcellation ). For the dHCP dataset, raw image scans are publicly available at https://nda.nih.gov/ . For the HCP dataset, raw image scans are publicly available at https://nda.nih.gov/ . Source data are provided with this paper. Code availability Software packages used in this manuscript include dHCP structural pipeline v1 ( https://github.com/BioMedIA/dhcp-structural-pipeline ), dHCP functional pipeline v1 ( https://git.fmrib.ox.ac.uk/seanf/dhcp-neonatal-fmri-pipeline ), SPM 12 ( http://www.fil.ion.ucl.ac.uk/spm/ ), GRETNA toolbox 2.0.0 ( https://www.nitrc.org/projects/gretna ), Individual Parcellation method ( https://github.com/hmlicas/Collaborative_Brain_Decomposition ), BrainSMASH ( https://github.com/murraylab/brainsmash/ ), LIBSVM (3.25) ( https://www.csie.ntu.edu.tw/∼cjlin/libsvm/ ), Support Vector Regression ( https://github.com/ZaixuCui/Pattern_Regression_Clean ), Support Vector Classification( https://github.com/ZaixuCui/Pattern_Classification ), cifti-matlab toolbox v2 ( https://github.com/Washington-University/cifti-matlab ), BrainNet Viewer 1.7 ( https://www.nitrc.org/projects/bnv ), R 4.0.3 ( https://www.r-project.org ), Matlab 2020b ( https://www.mathworks.com/products/matlab.html ), and Python 3.7.0 ( https://www.python.org ), online tools. The codes used in this study are available at github: https://github.com/zhaohuaxishi1/neonatal_Individual_Parcellation . Author Contributions J.L.Z., T.D.Z., and Y.H. designed research; T.D.Z., Y.H.X., H.M.L, Q.L.L, L.L.S., X.Y.L., M.Z.H., Z.L.Z., Z.X.C, and Y.H. provided the methodological instruction; J.L.Z. and T.D.Z. performed the data analysis; J.L.Z. and T.D.Z. developed visualizations. J.L.Z., T.D.Z., Z.X.C, and Y.H. wrote the paper; J.L.Z., T.D.Z., Z.X.C, and Y.H. revised the paper. Conflicting Interests The authors have declared that no conflicting interests exist. Acknowledgments This work was supported by the National Natural Science Foundation of China (Nos. 82021004, 31830034, 82102131, 82327807), Changjiang Scholar Professorship Award (No. T2015027), and the Fundamental Research Funds for the Central Universities (Nos. 2233300002, 2233100018). Funder Information Declared National Natural Science Foundation of China , 82021004 , 31830034 , 82102131 , 82327807 Changjiang Scholar Professorship Award , T2015027 Fundamental Research Funds for the Central Universities , 2233300002 , 2233100018 Footnotes https://github.com/zhaohuaxishi1/neonatal_Individual_Parcellation References 1. ↵ Gilmore , J.H. , Knickmeyer , R.C. & Gao , W . Imaging structural and functional brain development in early childhood . Nature Reviews Neuroscience 19 , 123 – 137 ( 2018 ). OpenUrl CrossRef PubMed 2. ↵ Cao , M. , Huang , H. & He , Y . Developmental connectomics from infancy through early childhood . Trends in neurosciences 40 , 494 – 506 ( 2017 ). OpenUrl CrossRef PubMed 3. ↵ Zhao , T. , Xu , Y. & He , Y . Graph theoretical modeling of baby brain networks . NeuroImage 185 , 711 – 727 ( 2019 ). OpenUrl CrossRef PubMed 4. ↵ Keunen , K. , Counsell , S.J. & Benders , M.J.N.L . The emergence of functional architecture during early brain development . NeuroImage 160 , 2 – 14 ( 2017 ). OpenUrl CrossRef PubMed 5. ↵ Gao , W . A hierarchical model of early brain functional network development . Trends in Cognitive Sciences 29 , 855 – 868 ( 2025 ). OpenUrl PubMed 6. ↵ Brady , R.G. , et al. Newborn Brain Function and Early Emerging Callous-Unemotional Traits . JAMA Psychiatry 81 , 303 – 311 ( 2024 ). OpenUrl PubMed 7. ↵ Sun , H. , et al. Brain age prediction and deviations from normative trajectories in the neonatal connectome . Nature Communications 15 , 10251 ( 2024 ). OpenUrl PubMed 8. ↵ Marín , O . Developmental timing and critical windows for the treatment of psychiatric disorders . Nature Medicine 22 , 1229 – 1238 ( 2016 ). OpenUrl CrossRef PubMed 9. Tooley , U.A. , et al. Prenatal environment is associated with the pace of cortical network development over the first three years of life . Nature Communications 15 , 7932 ( 2024 ). OpenUrl PubMed 10. ↵ Eyre , M. , et al. The Developing Human Connectome Project: typical and disrupted perinatal functional connectivity . Brain 144 , 2199 – 2213 ( 2021 ). OpenUrl CrossRef PubMed 11. ↵ Molloy , M.F. & Saygin , Z.M . Individual variability in functional organization of the neonatal brain . NeuroImage 253 , 119101 ( 2022 ). OpenUrl PubMed 12. ↵ Gao , W. , et al. Intersubject Variability of and Genetic Effects on the Brain’s Functional Connectivity during Infancy . The Journal of Neuroscience 34 , 11288 ( 2014 ). OpenUrl Abstract / FREE Full Text 13. ↵ Xu , Y. , et al. Development and Emergence of Individual Variability in the Functional Connectivity Architecture of the Preterm Human Brain . Cerebral Cortex 29 , 4208 – 4222 ( 2018 ). OpenUrl 14. ↵ Stoecklein , S. , et al. Variable functional connectivity architecture of the preterm human brain: Impact of developmental cortical expansion and maturation . Proceedings of the National Academy of Sciences 117 , 1201 – 1206 ( 2020 ). OpenUrl Abstract / FREE Full Text 15. ↵ Nielsen , A.N. , et al. Maturation of large-scale brain systems over the first month of life . Cerebral Cortex ( 2022 ). 16. ↵ Yin , W. , et al. The emergence of a functionally flexible brain during early infancy . Proceedings of the National Academy of Sciences 117 , 23904 – 23913 ( 2020 ). OpenUrl Abstract / FREE Full Text 17. ↵ Laumann , T.O. , et al. Functional system and areal organization of a highly sampled individual human brain . Neuron 87 , 657 – 670 ( 2015 ). OpenUrl CrossRef PubMed 18. Braga , R.M. , DiNicola , L.M. , Becker , H.C. & Buckner , R.L . Situating the left-lateralized language network in the broader organization of multiple specialized large-scale distributed networks . Journal of Neurophysiology 124 , 1415 – 1448 ( 2020 ). OpenUrl CrossRef PubMed 19. Gratton , C. , et al. Functional Brain Networks Are Dominated by Stable Group and Individual Factors, Not Cognitive or Daily Variation . Neuron 98 , 439 – 452 .e435 ( 2018 ). OpenUrl CrossRef PubMed 20. ↵ Wang , D. , et al. Parcellating cortical functional networks in individuals . Nature neuroscience 18 , 1853 – 1860 ( 2015 ). OpenUrl CrossRef PubMed 21. ↵ Kong , R. , et al. Spatial Topography of Individual-Specific Cortical Networks Predicts Human Cognition, Personality, and Emotion . Cerebral Cortex 29 , 2533 – 2551 ( 2018 ). OpenUrl 22. Lynch , C.J. , et al. Frontostriatal salience network expansion in individuals in depression . Nature 633 , 624 – 633 ( 2024 ). OpenUrl CrossRef PubMed 23. Du , J. , et al. Organization of the human cerebral cortex estimated within individuals: networks, global topography, and function . Journal of Neurophysiology 131 , 1014 – 1082 ( 2024 ). OpenUrl CrossRef PubMed 24. ↵ Gordon , E.M. , et al. A somato-cognitive action network alternates with effector regions in motor cortex . Nature 617 , 351 – 359 ( 2023 ). OpenUrl CrossRef PubMed 25. ↵ Cui , Z. , et al. Individual Variation in Functional Topography of Association Networks in Youth . Neuron 106 , 340 – 353 .e348 ( 2020 ). OpenUrl CrossRef PubMed 26. Hermosillo , R.J.M. , et al. A precision functional atlas of personalized network topography and probabilities . Nature Neuroscience 27 , 1000 – 1013 ( 2024 ). OpenUrl CrossRef PubMed 27. ↵ Keller , A.S. , et al. Personalized functional brain network topography is associated with individual differences in youth cognition . Nature Communications 14 , 8411 ( 2023 ). OpenUrl PubMed 28. ↵ Gordon , E.M. , Laumann , T.O. , Adeyemo , B. & Petersen , S.E . Individual variability of the system-level organization of the human brain . Cerebral cortex 27 , 386 – 399 ( 2017 ). OpenUrl CrossRef PubMed 29. ↵ Anderson , K.M. , et al. Heritability of individualized cortical network topography . Proceedings of the National Academy of Sciences 118 , e2016271118 ( 2021 ). OpenUrl Abstract / FREE Full Text 30. ↵ Dworetsky , A. , et al. Two common and distinct forms of variation in human functional brain networks . Nature Neuroscience 27 , 1187 – 1198 ( 2024 ). OpenUrl CrossRef PubMed 31. ↵ Edwards , A.D. , et al. The Developing Human Connectome Project Neonatal Data Release . Frontiers in Neuroscience 16 ( 2022 ). 32. ↵ Li , H. , Satterthwaite , T.D. & Fan , Y . Large-scale sparse functional networks from resting state fMRI . NeuroImage 156 , 1 – 13 ( 2017 ). OpenUrl CrossRef PubMed 33. ↵ Van Essen , D.C. , et al. The WU-Minn Human Connectome Project: An overview . NeuroImage 80 , 62 – 79 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 34. ↵ Yeo , B.T.T. , et al. The organization of the human cerebral cortex estimated by intrinsic functional connectivity . Journal of Neurophysiology 106 , 1125 – 1165 ( 2011 ). OpenUrl CrossRef PubMed Web of Science 35. ↵ Gordon , E.M. , et al. Precision functional mapping of individual human brains . Neuron 95 , 791 – 807 . e797 ( 2017 ). OpenUrl CrossRef PubMed 36. ↵ Doria , V. , et al. Emergence of resting state networks in the preterm human brain . Proceedings of the National Academy of Sciences 107 , 20015 – 20020 ( 2010 ). OpenUrl Abstract / FREE Full Text 37. ↵ Gao , W. , et al. Functional Network Development During the First Year: Relative Sequence and Socioeconomic Correlations . Cerebral Cortex 25 , 2919 – 2928 ( 2014 ). OpenUrl PubMed 38. ↵ Fransson , P. , et al. Resting-state networks in the infant brain . Proceedings of the National Academy of Sciences 104 , 15531 – 15536 ( 2007 ). OpenUrl Abstract / FREE Full Text 39. ↵ Gao , W. , et al. Evidence on the emergence of the brain’s default network from 2-week-old to 2-year-old healthy pediatric subjects . Proceedings of the National Academy of Sciences 106 , 6790 – 6795 ( 2009 ). OpenUrl Abstract / FREE Full Text 40. ↵ Leys , C. , Ley , C. , Klein , O. , Bernard , P. & Licata , L . Detecting outliers: Do not use standard deviation around the mean, use absolute deviation around the median . Journal of experimental social psychology 49 , 764 – 766 ( 2013 ). OpenUrl CrossRef 41. ↵ Li , M. , et al. Performing group-level functional image analyses based on homologous functional regions mapped in individuals . PLoS biology 17 , e2007032 ( 2019 ). OpenUrl CrossRef PubMed 42. ↵ Bayley , N . BSID-III: Bayley Scales of infant development . ( Firenze. Organizzazioni Speciali , 2006 ). 43. ↵ Smyser , C.D. , et al. Longitudinal Analysis of Neural Network Development in Preterm Infants . Cerebral Cortex 20 , 2852 – 2862 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 44. ↵ França , L.G.S. , et al. Neonatal brain dynamic functional connectivity in term and preterm infants and its association with early childhood neurodevelopment . Nature Communications 15 , 16 ( 2024 ). 45. ↵ Smith , S.M. , Vidaurre , D. , Alfaro-Almagro , F. , Nichols , T.E. & Miller , K.L . Estimation of brain age delta from brain imaging . NeuroImage 200 , 528 – 539 ( 2019 ). OpenUrl CrossRef PubMed 46. ↵ Tian , Y.E. , et al. Heterogeneous aging across multiple organ systems and prediction of chronic disease and mortality . Nature Medicine ( 2023 ). 47. ↵ Liu , S. , Abdellaoui , A. , Verweij , K.J.H. & van Wingen , G.A . Replicable brain–phenotype associations require large-scale neuroimaging data . Nature Human Behaviour ( 2023 ). 48. ↵ Marek , S. , et al. Reproducible brain-wide association studies require thousands of individuals . Nature 603 , 654 – 660 ( 2022 ). OpenUrl CrossRef PubMed 49. ↵ Smith , S.M. , et al. Correspondence of the brain’s functional architecture during activation and rest . Proceedings of the National Academy of Sciences 106 , 13040 – 13045 ( 2009 ). OpenUrl Abstract / FREE Full Text 50. ↵ Power , Jonathan D. , et al. Functional Network Organization of the Human Brain . Neuron 72 , 665 – 678 ( 2011 ). OpenUrl CrossRef PubMed Web of Science 51. ↵ Sylvester , C.M. , et al. Network-specific selectivity of functional connections in the neonatal brain . Cerebral Cortex 33 , 2200 – 2214 ( 2022 ). OpenUrl 52. ↵ Silbereis , John C. , Pochareddy , S. , Zhu , Y. , Li , M. & Sestan , N . The Cellular and Molecular Landscapes of the Developing Human Central Nervous System . Neuron 89 , 248 – 268 ( 2016 ). OpenUrl CrossRef PubMed 53. ↵ Wu , Y.J. , et al. Rapid learning of a phonemic discrimination in the first hours of life . Nature Human Behaviour 6 , 1169 – 1179 ( 2022 ). OpenUrl PubMed 54. ↵ Gao , W. , Alcauter , S. , Smith , J.K. , Gilmore , J.H. & Lin , W . Development of human brain cortical network architecture during infancy . Brain Structure and Function 220 , 1173 – 1186 ( 2015 ). OpenUrl 55. ↵ Myers , M.J. , et al. Functional parcellation of the neonatal cortical surface . Cerebral Cortex 34 ( 2024 ). 56. ↵ Moore , L.A. , et al. Towards personalized precision functional mapping in infancy . Imaging Neuroscience 2 , 1 – 20 ( 2024 ). OpenUrl CrossRef 57. ↵ Buckner , R.L. & Krienen , F.M . The evolution of distributed association networks in the human brain . Trends in Cognitive Sciences 17 , 648 – 665 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 58. ↵ Sydnor , V.J. , et al. Neurodevelopment of the association cortices: Patterns, mechanisms, and implications for psychopathology . Neuron 109 , 2820 – 2846 ( 2021 ). OpenUrl CrossRef PubMed 59. ↵ Du , J. , et al. Organization of the human cerebral cortex estimated within individuals: networks, global topography, and function . Journal of neurophysiology 131 , 1014 – 1082 ( 2024 ). OpenUrl CrossRef PubMed 60. ↵ Gao , W. , et al. A review on neuroimaging studies of genetic and environmental influences on early brain development . NeuroImage 185 , 802 – 812 ( 2019 ). OpenUrl CrossRef PubMed 61. ↵ Dosenbach , N.U.F. , et al. Prediction of Individual Brain Maturity Using fMRI . Science 329 , 1358 – 1361 ( 2010 ). OpenUrl Abstract / FREE Full Text 62. ↵ Scheinost , D. , et al. Machine learning and prediction in fetal, infant, and toddler neuroimaging: a review and primer . Biological Psychiatry ( 2022 ). 63. ↵ Smyser , C.D. , et al. Prediction of brain maturity in infants using machine-learning algorithms . NeuroImage 136 , 1 – 9 ( 2016 ). OpenUrl PubMed 64. ↵ Cao , M. , et al. Early Development of Functional Network Segregation Revealed by Connectomic Analysis of the Preterm Human Brain . Cerebral Cortex 27 , 1949 – 1963 ( 2016 ). OpenUrl 65. ↵ Dehaene-Lambertz , G. & Spelke , E.S . The Infancy of the Human Brain . Neuron 88 , 93 – 109 ( 2015 ). OpenUrl CrossRef PubMed 66. ↵ Baker , S. & Kandasamy , Y . Machine learning for understanding and predicting neurodevelopmental outcomes in premature infants: a systematic review . Pediatric Research 93 , 293 – 299 ( 2023 ). OpenUrl PubMed 67. ↵ Ouyang , M. , et al. Diffusion-MRI-based regional cortical microstructure at birth for predicting neurodevelopmental outcomes of 2-year-olds . Elife 9 , e58116 ( 2020 ). OpenUrl CrossRef PubMed 68. ↵ Sun , H. , et al. Network controllability of structural connectomes in the neonatal brain . Nature Communications 14 , 5820 ( 2023 ). OpenUrl PubMed 69. ↵ Hu , D. , et al. Existence of Functional Connectome Fingerprint during Infancy and Its Stability over Months . The Journal of Neuroscience 42 , 377 – 389 ( 2022 ). OpenUrl Abstract / FREE Full Text 70. ↵ Wang , Q. , et al. Individual Uniqueness in the Neonatal Functional Connectome . Cerebral Cortex 31 , 3701 – 3712 ( 2021 ). OpenUrl CrossRef PubMed 71. ↵ Xia , Y. , et al. Development of sensorimotor-visual connectome gradient at birth predicts neurocognitive outcomes at 2 years of age . iScience 27 , 108981 ( 2024 ). OpenUrl PubMed 72. ↵ Johnson , M.H . Sensitive periods in functional brain development: problems and prospects . Developmental Psychobiology: The Journal of the International Society for Developmental Psychobiology 46 , 287 – 292 ( 2005 ). OpenUrl 73. ↵ Mesulam , M.-M . From sensation to cognition . Brain: a journal of neurology 121 , 1013 – 1052 ( 1998 ). OpenUrl CrossRef PubMed Web of Science 74. ↵ Dosenbach , N.U.F. , Raichle , M.E. & Gordon , E.M . The brain’s action-mode network . Nature Reviews Neuroscience 26 , 158 – 168 ( 2025 ). OpenUrl CrossRef PubMed 75. ↵ Scharf , R.J. , Scharf , G.J. & Stroustrup , A. Developmental Milestones . Pediatrics In Review 37 , 25 – 38 ( 2016 ). OpenUrl PubMed 76. ↵ Nelson , C.A. , et al. Cognitive Recovery in Socially Deprived Young Children: The Bucharest Early Intervention Project . Science 318 , 1937 – 1940 ( 2007 ). OpenUrl Abstract / FREE Full Text 77. ↵ Grossmann , T . The role of medial prefrontal cortex in early social cognition . Frontiers in Human Neuroscience 7 ( 2013 ). 78. ↵ Perani , D. , et al. Functional specializations for music processing in the human newborn brain . Proceedings of the National Academy of Sciences 107 , 4758 – 4763 ( 2010 ). OpenUrl Abstract / FREE Full Text 79. ↵ Shine , James M. , et al. The Dynamics of Functional Brain Networks: Integrated Network States during Cognitive Task Performance . Neuron 92 , 544 – 554 ( 2016 ). OpenUrl CrossRef PubMed 80. ↵ Ball , G. , et al. Development of cortical microstructure in the preterm human brain . Proceedings of the National Academy of Sciences 110 , 9541 – 9546 ( 2013 ). OpenUrl Abstract / FREE Full Text 81. ↵ Giménez , M. , et al. Accelerated cerebral white matter development in preterm infants: A voxel-based morphometry study with diffusion tensor MR imaging . NeuroImage 41 , 728 – 734 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 82. ↵ Smyser , C.D. , et al. Resting-State Network Complexity and Magnitude Are Reduced in Prematurely Born Infants . Cerebral Cortex 26 , 322 – 333 ( 2014 ). OpenUrl PubMed 83. ↵ De Asis-Cruz , J. , et al. Functional brain connectivity in ex utero premature infants compared to in utero fetuses . NeuroImage 219 , 117043 ( 2020 ). OpenUrl CrossRef PubMed 84. ↵ Lee , Joo Y. , Park , H.-K. & Lee , Hyun J . Accelerated Small-World Property of Structural Brain Networks in Preterm Infants at Term-Equivalent Age . Neonatology 115 , 99 – 107 ( 2018 ). OpenUrl PubMed 85. ↵ Pittet , M.P. , Vasung , L. , Huppi , P.S. & Merlini , L . Newborns and preterm infants at term equivalent age: A semi-quantitative assessment of cerebral maturity . NeuroImage: Clinical 24 , 102014 ( 2019 ). OpenUrl PubMed 86. ↵ Kelly , C. , et al. Investigating brain structural maturation in children and adolescents born very preterm using the brain age framework . NeuroImage 247 , 118828 ( 2022 ). OpenUrl PubMed 87. ↵ Toda , T. , et al. Birth Regulates the Initiation of Sensory Map Formation through Serotonin Signaling . Developmental Cell 27 , 32 – 46 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 88. ↵ Witteveen , I.F. , et al. Preterm birth accelerates the maturation of spontaneous and resting activity in the visual cortex . Frontiers in Integrative Neuroscience 17 ( 2023 ). 89. ↵ Yin , W. , et al. Charting brain functional development from birth to 6 years of age . Nature Human Behaviour , 1 – 14 ( 2025 ). 90. ↵ Korom , M. , et al. Dear reviewers: Responses to common reviewer critiques about infant neuroimaging studies . Dev Cogn Neurosci 53 , 101055 ( 2022 ). OpenUrl PubMed 91. ↵ Thapar , A. , Cooper , M. & Rutter , M. Neurodevelopmental disorders . The Lancet Psychiatry 4 , 339 – 346 ( 2017 ). OpenUrl PubMed 92. ↵ Allison , C. , et al. The Q-CHAT (Quantitative CHecklist for Autism in Toddlers): A Normally Distributed Quantitative Measure of Autistic Traits at 18–24 Months of Age: Preliminary Report . Journal of Autism and Developmental Disorders 38 , 1414 – 1425 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 93. ↵ Makropoulos , A. , et al. The developing human connectome project: A minimal processing pipeline for neonatal cortical surface reconstruction . NeuroImage 173 , 88 – 112 ( 2018 ). OpenUrl CrossRef PubMed 94. ↵ Serag , A. , et al. Construction of a consistent high-definition spatio-temporal atlas of the developing brain using adaptive kernel regression . NeuroImage 59 , 2255 – 2265 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 95. ↵ Yan , C.-G. , et al. A comprehensive assessment of regional variation in the impact of head micromovements on functional connectomics . NeuroImage 76 , 183 – 201 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 96. ↵ Power , J.D. , Barnes , K.A. , Snyder , A.Z. , Schlaggar , B.L. & Petersen , S.E . Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion . NeuroImage 59 , 2142 – 2154 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 97. ↵ Glasser , M.F. , et al. The minimal preprocessing pipelines for the Human Connectome Project . NeuroImage 80 , 105 – 124 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 98. ↵ Burt , J.B. , Helmer , M. , Shinn , M. , Anticevic , A. & Murray , J.D . Generative modeling of brain maps with spatial autocorrelation . NeuroImage 220 , 117038 ( 2020 ). OpenUrl CrossRef PubMed 99. ↵ Eklund , A. , Nichols , T.E. & Knutsson , H . Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates . Proceedings of the National Academy of Sciences 113 , 7900 – 7905 ( 2016 ). OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted November 19, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Emergence of Individualized Functional Topography in the Neonatal Brain Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Emergence of Individualized Functional Topography in the Neonatal Brain Jianlong Zhao , Yuehua Xu , Zaixu Cui , Hongming Li , Lianglong Sun , Xinyuan Liang , Meizhen Han , Zilong Zeng , Qiongling Li , Tengda Zhao , Yong He bioRxiv 2025.11.18.689151; doi: https://doi.org/10.1101/2025.11.18.689151 Share This Article: Copy Citation Tools Emergence of Individualized Functional Topography in the Neonatal Brain Jianlong Zhao , Yuehua Xu , Zaixu Cui , Hongming Li , Lianglong Sun , Xinyuan Liang , Meizhen Han , Zilong Zeng , Qiongling Li , Tengda Zhao , Yong He bioRxiv 2025.11.18.689151; doi: https://doi.org/10.1101/2025.11.18.689151 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17697) Bioengineering (13894) Bioinformatics (41951) Biophysics (21455) Cancer Biology (18592) Cell Biology (25507) Clinical Trials (138) Developmental Biology (13380) Ecology (19903) Epidemiology (2067) Evolutionary Biology (24321) Genetics (15610) Genomics (22509) Immunology (17737) Microbiology (40398) Molecular Biology (17182) Neuroscience (88618) Paleontology (667) Pathology (2833) Pharmacology and Toxicology (4825) Physiology (7641) Plant Biology (15158) Scientific Communication and Education (2046) Synthetic Biology (4296) Systems Biology (9825) Zoology (2271)