Bin Wan

Neurotransmitter receptor and transporter distributions shape neural communication across the cortex and underpin mechanistic interpretations of brain function. Magnetic resonance imaging (MRI)-derived eigenmodes, low-dimensional spatial patterns that capture the principal axes of cortical organization, are increasingly used to reconstruct brain maps and offer a potential route to approximating molecular topographies in large cohorts where direct measurements are unavailable. Yet whether the type of MRI data used to derive these eigenmodes determines which receptor and transporter distributions can be approximated, and which molecular systems are accessible through this approach at all, remains unknown. Using a spatial reconstruction framework applied to 23 molecular systems spanning neurotransmitter, metabolic, and pathological markers, we show that eight receptor and transporter maps, spanning serotonin, dopamine, histamine, cholinergic, and opioid markers, can be selectively approximated from functional eigenmodes but not structural or geometric ones. This selectivity also reveals that functional brain organization encodes a specific subset of neuromodulatory topographies that structural and geometric organization does not. To illustrate translational potential in clinical cohorts where molecular imaging is unavailable, we applied individual-level eigenmode decompositions of MRI data projected onto the eight validated molecular systems and detected cross-sectional differences in autism and longitudinal changes in schizophrenia. Together, these findings reveal a difference in how geometric, structural, and functional brain organization relate to molecular architecture, offering a method for approximating validated molecular profiles in population neuroimaging cohorts that lack direct molecular imaging.

Background: Although most individuals with schizophrenia do not show aggressive behavior, rates are higher than in the general population. Identifying the neurobiological correlates of aggression may inform prevention. Methods: We included 2260 individuals with schizophrenia spectrum disorders and 3058 healthy controls (HC) from 26 sites in the ENIGMA-Schizophrenia Working Group. The former were classified as showing at least moderate aggression (SZ+A; N = 231) or not (SZ−A; N = 2029) based on symptom rating scales. Structural covariance networks were constructed from cortical and subcortical volumes. We examined global and nodal network topology, as well as density within six large-scale brain networks and between each network and the rest of the brain. Metrics were summarized as the area under the curve (AUC) across a prespecified range of densities ( K = 0.10–0.25, Δ K = 0.01). Group differences were assessed using pairwise permutation tests with Benjamini–Hochberg correction for multiple comparisons. Results: Global and nodal network topology did not differ significantly between SZ+A and HC. Network–rest density of the salience network (SN) was higher in SZ+A than in HC, although this difference did not survive multiple-comparison correction (ΔAUC = 0.018, p _BH = .083). In exploratory subgroup analyses, higher SN network–rest density was observed only in SZ+A with high delusion severity (ΔAUC = 0.021, p = .007). By contrast, SZ−A demonstrated extensive alterations in global network topology compared with SZ+A and HC, alongside selective alterations in nodal topology and large-scale network density compared with HC. Results were broadly consistent across sensitivity analyses using alternative definitions of aggression and adjusting for current antipsychotic dose. Conclusions: SZ+A was characterized by fewer—not more—abnormalities in structural covariance network organization than SZ−A, which may reflect etiological heterogeneity. In SZ+A with high delusion severity, higher network–rest density of the SN may point to aberrant salience processing in aggressive behavior.

Brain network modeling uses computer simulations to infer about latent neural properties at micro- and mesoscales by fitting brain dynamic models to empirical data of individual subjects or groups. However, computational costs of (individualized) model fitting is a major bottleneck, limiting the practical feasibility of this approach to larger cohorts and more complex models, and highlighting the need for scalable simulation implementations. Here, we introduce cuBNM, a Python package which leverages parallel processing of graphics processing units to massively accelerate simulations of brain network models. We show running simulations on graphics processing units is several hundred times faster compared to central processing units. We demonstrate the usage of cuBNM by running optimization of group-level and individualized low- and high-dimensional models. As examples of the utility of individualized models, we investigated test-retest reliability and heritability of simulated and empirical measures in the Human Connectome Project dataset. We found simulated features were fairly reliable and significantly heritable, suggesting their biological plausibility. Overall, cuBNM enables large-scale simulations of brain network models, opening new avenues for studying latent neural processes across diverse populations, dense networks, and high-dimensional models, which was previously impractical due to computational constraints.

The hippocampus and amygdala are neighboring medial temporal lobe structures linked to memory and affect, yet how their subregions are jointly embedded within distributed isocortical systems remains unclear. Using resting-state fMRI from 722 Human Connectome Project Young Adult participants, we mapped hippocampal and amygdalar subregions within a unified cortex-wide framework, quantifying subregion-to-cortex connectivity via Pearson correlation (broad co-fluctuation) and GLASSO partial correlation (relatively more direct functional association). We introduced two count-based metrics: dominance (relative hippocampal vs. amygdalar representation) and sharedness (balanced co-representation). Direct associations showed both structures sharing coupling with paralimbic areas and, more modestly, default mode regions, while broader co-fluctuations extended into somatomotor and paralimbic networks. Divergence patterns depended on the estimator: hippocampal subregions preferentially coupled with default-mode and visual networks under direct association, while amygdalar nuclei favored ventral attention and limbic networks; broader co-fluctuations additionally implicated somatomotor cortex for amygdala and visual cortex for hippocampus. These principles held at the subfield/nucleus level, varying along the hippocampal long axis and identifying the paralaminar nucleus as the most hippocampus-like amygdalar subregion. Data-driven connectivity gradients confirmed both systems’ separation and fine-scale interdigitation. Hippocampal and amygdalar subregions are thus embedded in cortex not as discrete systems, but through structured, spatially organized co-representation.

Epilepsy is characterized by widespread structural brain alterations extending beyond the epileptic zone, involving both cortical and subcortical regions. Importantly, the clinical manifestation of epilepsy, including seizure types, psychiatric comorbidities, and treatment responses, has been shown to differ between sexes. However, sex differences in structural alterations in epilepsy have been seldomly reported in neuroimaging studies, partly due to limited sample sizes and single-center designs. Here, we systematically investigated sex differences in common epilepsies and their related clinical variables using structural neuroimaging biomarkers in an international multi-center cohort of 1,253 epilepsy patients and 1,077 healthy controls. We studied cortical thickness and subcortical volume in two types of epilepsy: temporal lobe epilepsy (TLE) and genetic generalized epilepsy (GGE). Both male and female patients with TLE showed widespread cortical and subcortical thinning compared with controls. In GGE, when compared separately to controls, male patients showed only subtle structural alterations, whereas female patients exhibited more widespread structural alterations. Sex-stratified analyses revealed some variation in the extent and distribution of cortical thickness and subcortical volume alterations between male and female patients in both epilepsy cohorts. Yet, we did not find significant sex-by-diagnosis interaction effects in TLE and GGE. Similarly, no significant interaction effects were observed between sex and age of onset or disease duration in either patient group. Overall, although we observed some differences in regional cortical thickness and subcortical volume between male and female patients with epilepsy, we did not find significant sex-by-diagnosis interactions. Our findings indicate that sex differences in behavioral and clinical outcomes of epilepsy may involve biological or functional processes that require further investigation.

Stroke starts as a focal vascular lesion, but its structural consequences often extend beyond the lesion site, resulting in distributed brain atrophy whose organizing principles remain unclear. Here, using longitudinal MRI data from two stroke cohorts spanning the hyperacute to chronic stages, we tested whether and how post-stroke grey matter volume (GMV) change over time is constrained by large-scale brain networks across individuals. Coordinated deformation modelling showed that longitudinal GMV atrophy patterns were more strongly shaped by structural than by functional connectivity, with structural constraints already detectable in the acute phase. Network diffusion modelling further showed that atrophy patterns at 3 and 12 months were consistent with lesion-initiated propagation on the structural connectome, whereas model performance was weak in the early acute stage. The model-derived propagation stage did not increase monotonically with chronological time, indicating that it reflects the degree to which observed atrophy conforms to a network-consistent diffusion-like pattern rather than elapsed biological time. Inter-individual variation in propagation stage was associated with lesion size and with the topological and molecular properties of lesioned regions. Finally, lesion information combined with network diffusion modelling supported individualized prediction of later GMV atrophy patterns. These findings provide a network-based account of secondary degeneration after stroke and may inform individualized characterization of post-stroke structural reorganization.

Schizophrenia is often conceptualized as a brain network disorder, yet the organizational principles and heterogeneity underlying widespread cortical abnormalities remain poorly understood. Leveraging multisite MRI data from 3,958 individuals diagnosed with schizophrenia and 5,489 neurotypical individuals, we studied the cortical organization and its subtyping by analyzing individualized cortical network similarity. We used eigenvector decompositions to study spatial patterning of the gradients and graph theory to study small–world topology. Individuals with schizophrenia showed widespread alterations of gradient loadings, which followed inferior–superior and frontal–temporal axes. Alterations in small–world topology were localized in key network hubs, including the insula and anterior cingulate cortex. Brain–symptom association analyses identified a latent dimension linking disorganization symptoms to topological alterations. Finally, clustering cortical alterations identified two robust subtypes, characterized by divergent anterior cingulate (S1) versus temporoparietal (S2) thickness differences aligned with the intrinsic gradient-topology patterns. Both subtypes were present early in the illness and stable across disease stages and age groups. These findings reveal systematic disruptions of cortical organization in schizophrenia, providing a network-level framework for macroscale brain organization and inter-individual heterogeneity.

The human brain requires a continuous supply of energy to function effectively. Here, we investigated how the low-dimensional organization of intrinsic functional connectivity patterns based on resting-state functional magnetic resonance imaging relates to brain energy expenditure measured by fluorodeoxyglucose positron emission tomography. By incrementally adding more dimensions of brain organization (via functional gradients), we show that increasing amounts of variance in the map of brain energy expenditure are accounted for. Dimensions of brain organization that explained much of the variance in intrinsic brain function also accounted for a substantial share of regional variance in energy expenditure maps. This relationship was especially pronounced for maps based on the strongest connections, suggesting that weaker connections may contribute less to explaining regional energy variance. Notably, our topological model was more effective than random brain organization configurations, suggesting that brain organization may be specifically associated with energy optimization. Our results demonstrate how the spatial organization of functional connections is systematically linked to optimized energy expenditure in the human brain, providing new insights into the metabolic basis of brain function.

Brain network modeling uses computer simulations to infer about latent neural properties at micro- and mesoscales by fitting brain dynamic models to empirical data of individual subjects or groups. However, computational costs of (individualized) model fitting is a major bottleneck, limiting the practical feasibility of this approach to larger cohorts and more complex models, and highlighting the need for scalable simulation implementations. Here, we introduce cuBNM, a Python package which leverages parallel processing of graphics processing units to massively accelerate simulations of brain network models. We show running simulations on graphics processing units is several hundred times faster compared to central processing units. We demonstrate the usage of cuBNM by running optimization of group-level and individualized low- and high-dimensional models. As examples of the utility of individualized models, we investigated test-retest reliability and heritability of simulated and empirical measures in the Human Connectome Project dataset. We found simulated features were fairly reliable and significantly heritable, suggesting their biological plausibility. Overall, cuBNM enables large-scale simulations of brain network models, opening new avenues for studying latent neural processes across diverse populations, dense networks, and high-dimensional models, which was previously impractical due to computational constraints.

Schizophrenia is a neurodevelopmental condition with alterations in both sensory and association cortical areas. These alterations have been reported to follow structural connectivity patterning, and to occur in a system-level fashion. Here we investigated whether pathological alterations of schizophrenia originate from an early disruption of cortical organization. We found a structural covariance gradient axis of cortical thickness discriminated anterior from posterior region and was compressed in early-onset schizophrenia (EOS) patients. Patients showed increased structural covariance between two ends of the anterior-posterior axis, with increased geodesic distance of covarying regions between two ends. Positive symptoms increased with the strengthening of structural covariance between two ends. Our findings revealed a contracted organizational axis in EOS patients, which was attributed to excessive distally coordinated changes between anterior and posterior cortical regions. Our study from a systematic perspective suggests disturbed maturational processes of cortical thickness in EOS, supporting the neurodevelopmental hypothesis of schizophrenia.