Armin Iraji

Psychosis exhibits marked neurobiological heterogeneity that is not adequately captured by symptom-based diagnoses. Although prior efforts have identified subgroups, it remains unclear how latent disease variability manifests within the brain’s multiscale intrinsic functional organization and whether these patterns can be leveraged to systematically identify biologically driven subtypes of psychosis. Here, we aim to identify Psychosis Imaging Neurosubtypes (PINs)—homogeneous subgroups defined solely from neurobiological features of brain function, independent of clinical or cognitive definitions, and whether these subgroups show meaningful phenotypic differentiation. Resting-state fMRI data from 2103 B-SNIP 1&2 participants were used to compute subject-specific multiscale functional network connectivity (msFNC). We then derived a low-dimensional neurobiological subspace, termed Latent Network Connectivity, which captured system-wide interconnected multiscale information across three components. Projections of probands msFNC onto this subspace revealed three PINs through unsupervised learning that spanned DSM diagnoses (Schizophrenia, Bipolar, Schizoaffective Disorder). Although defined purely from intrinsic neurobiology, PINs exhibited distinct cognitive, clinical, and connectivity profiles. PIN-1, the most cognitively impaired, showed Cerebellar-Subcortical and Visual-Sensorimotor hypoconnectivity, alongside Visual-Subcortical hyperconnectivity. Most cognitively preserved PIN-2 showed Visual-Subcortical, Subcortical-Sensorimotor, and Subcortical-Higher Cognition hypoconnectivity. PIN-3 exhibited intermediate cognition, showing Cerebellar-Subcortical hypoconnectivity alongside Cerebellar-Sensorimotor and Subcortical-Sensorimotor hyperconnectivity. Although covariate-adjusted analyses did not reveal significant clinical differences among PINs, exploratory analyses identified selective differences in YMDR, PANSS, and Age of Onset. Notably, 55% of relatives aligned with the same neurosubtype as their probands, a rate significantly higher than chance and than DSM-based alignment, suggesting potential heritability. Cognitive performance reliably aligns with distinct brain connectivity patterns, which are also evident in relatives, supporting their construct validity. Our PINs partially converged with previously reported BSNIP Biotypes with some correspondence (e.g. PIN1 with Biotype 1) but not with the other Biotypes, which were determined using integrated electrophysiological, cognitive, and oculomotor data, supporting convergent validity, while also indicating that intrinsic rsfMRI derived multiscale connectivity partitions psychosis heterogeneity somewhat differently from multimodal biomarker approaches. These findings underscore the limitations of DSM-based classifications in capturing the biological complexity of psychotic disorders and highlight the potential of imaging-based neurosubtypes to enhance our understanding of the psychosis spectrum.

The biological heterogeneity of psychotic disorders (PDs) has motivated the identification of psychosis imaging neurosubtypes (PINs), which aims to stratify the PD population into subgroups with similar neurobiological underpinnings. Yet the influence of sex, although well documented in the literature, has not been well counted in the existing subtyping paradigm. Here, we introduce concurrent supervised-unsupervised representative subspace clustering (CSU-RSC), a general functional neurosubtyping framework that jointly estimates cluster-specific subspaces and a partition while softly incorporating a supervised label, aiming to identify sex-dominant subgroups based on differences in the spatial organization of brain networks. This soft supervising design differentiates CSU-RSC from unsupervised approaches that completely ignore a label as well as from supervised approaches that exclusively analyze each label group separately. On simulated data, CSU-RSC recovered the ground truth of subgroups more accurately than its unsupervised counterpart. Applied to resting-state fMRI from 1,239 individuals with psychosis, CSU-RSC identified two sex-dominant PINs that replicated across discovery and validation sets, and showed significant differences in clinical characteristics, including cognitive impairment and symptom severity, with the most cognitively impaired subtype being female-dominant. Projecting controls onto the patient-derived subspaces showed that the female-dominant pattern did not persist in controls, suggesting psychosis-specific sex heterogeneity. Overall, by incorporating sex as an explicit source of information, CSU-RSC highlights the value of sex for resolving biological heterogeneity and provides a basis for more reproducible and biologically informative patient stratification.

Psychotic disorders are severe mental conditions whose biological underpinnings remain elusive. This is in part due to marked heterogeneity, which may have precluded advances in precise understanding and treatment of the disorders. Previous work has tried to elucidate this heterogeneity; nevertheless, sex, as an important biological variable, has not been well accounted for despite established sex differences. Using resting-state fMRI data from the BSNIP consortium (N=1753; 64.3% probands with psychosis include 571 females and 556 males; 35.7% of controls includes 369 females and 257 males), we extracted multiscale functional network connectivity and employed unsupervised learning methods to identify potential subgroups with distinct neurobiological profiles, characterized by an overrepresentation of males or females (termed here as sex-dominant Psychosis Imaging Neurosubtypes, PINs). Four sex-dominant neurosubtypes emerged, two female-dominant (fPIN-1, fPIN-2) and two male-dominant (mPIN-1, mPIN-2). Each PIN exhibited distinct, replicable dysconnectivity patterns that uniquely contributed to cognitive deficits. Brain-Predicted Cognitive performance derived from these PIN-specific dysconnectivity, prominent among the triple network (default mode and salience), subcortical-basal ganglia, higher cognition-insular temporal, and visual-occipitotemporal subdomains, correlated with measured cognitive performance and showed PIN-specific reductions relative to controls, confirming neurosubtype-congruent associations. Moreover, fPIN-2's distinct dysconnectivity also predicted positive and negative symptom scores. Our findings lend support for sex-dominant neurobiological heterogeneity, with sex-dominant PINs presenting unique dysconnectivity patterns that contribute to cognitive and symptom outcomes, even though neurosubtypes appear clinically similar. These results establish sex as a critical biological variable in deconstructing psychosis heterogeneity, which holds promise in revealing more precise biomarkers for disease characterization and guiding personalized treatment strategies.

Early infancy is characterized by rapid, nonlinear reorganization of the functional connectome, yet its dominant growth modes and relationship to short-timescale brain dynamics remain unclear. Using a densely sampled longitudinal multimodal dataset collected across the first six postnatal months, we estimated days-scale trajectories of whole-brain functional connectivity growth and visually guided behaviors. We identify a growth-mode transition in the functional connectome at approximately 2-4 months, marked by a shift from predominantly local strengthening toward increasingly distributed integration. This developmental reorganization was preceded by changes in the fluctuation asymmetry ratio (FLARe), a measure of asymmetry in recurring short-timescale co-fluctuation states. Across individuals, FLARe predicted subsequent change rates of functional connectivity and visually guided behaviors. These findings identify short-timescale fluctuation asymmetry as a previously unrecognized prospective signal of later days-scale connectome and behavioral change, and provide a cross-timescale framework for linking functional brain dynamics unfolding over seconds to developmental change emerging over weeks to months.

The integration of EEG and fMRI offers a powerful method for exploring the brain's spatial and temporal dynamics. However, existing approaches typically summarize both EEG and fMRI, often removing temporal information before combining the modalities. Our novel approach brings together group ICA of fMRI, group ICA of EEG, and joint ICA to propose a multimodal data fusion approach, named group joint ICA (gjICA), that links simultaneous EEG‐fMRI data from multiple subjects. The proposed framework enables group‐level mapping and provides single‐subject estimates via back‐reconstruction, facilitating a comprehensive picture of brain networks. The gjICA also introduces joint functional network connectivity (jFNC), which provides connectivity between fMRI networks as well as between EEG components, hence linking temporal and spatial information. When applied to concurrent resting EEG‐fMRI data from 121 participants, we identified 63 multimodal brain components. The statistical analyses of these components further revealed that depression is characterized by widespread multimodal connectivity, with visual and higher cognition networks acting as hubs in these alterations. Further, joint histogram analysis revealed that depression is broadly associated with reduced EEG component expression with increased fMRI component expression in cerebellar, visual, subcortical, and sensorimotor networks, suggesting altered neurovascular coupling and modality‐specific dysfunctions. Overall, the proposed gjICA approach provides a robust framework for fusing EEG and fMRI data while preserving the spatiotemporal information in both modalities, enabling the identification of a more comprehensive picture of multimodal neural relationships, enhancing our understanding of brain dynamics and providing new insights into complex brain disorders such as depression.

Nonlinearity is a hallmark of brain complexity at multiple scales. However, existing functional magnetic resonance imaging (fMRI) functional connectivity studies typically utilize linear methods. Therefore, links between nonlinear fMRI connectivity patterns and the development of the human brain during critical periods such as infancy remain unclear. To address this gap in knowledge, we developed a data-driven approach to capture brain intrinsic connectivity networks from explicitly nonlinear resting-state fMRI connectivity and profiled their developmental associations in a cohort of typically developing human infants. We identified neurobiologically structured nonlinear fMRI connectivity patterns during early postnatal life, indicating that macroscopic brain ensembles systematically participate in nonlinear relationships at birth. Furthermore, we found that linear and explicitly nonlinear network counterparts are linked to partially overlapping but complementary developmental profiles during this period of rapid brain maturation, with the explicitly nonlinear approach unveiling insights into the development of networks that have been associated with sensorimotor capacities, default mode processes, executive functioning, language production, and stimulus saliency. Our study marks the first comprehensive developmental investigation of whole-brain nonlinear fMRI networks in human infants and deepens contemporary perspectives on neuroimaging data discovery by emphasizing the informational richness of nonlinear relationships at fMRI scales of observation.

The brain generates diverse cognitive states while maintaining a stable functional architecture, a duality that remains difficult to reconcile. Prevailing views assume that flexible cognition necessitates correspondingly flexible architecture, in which different tasks demand distinct reconfigurations of functional networks. Here we introduce the intrinsic network flow (INF) framework as a complementary view. This framework is built on temporally coordinated signal flows across brain networks that constitute a universal scaffold, stable across diverse cognitive states and common across individuals. We show that a wide range of task-evoked activation patterns can be reconstructed by modulating only the temporal phase alignment of these flows, whose fixed structure determines functional connectivity topology, gradients, and large-scale networks, thereby preserving these properties across task states. This situates resting-state and task-state dynamics within a unified framework and suggests a generative relationship from flow-like dynamics to the full landscape of resting-state and task-state phenomena. Crucially, phase information, which neither existing brain state analyses nor eigenmode decompositions can extract, outperforms amplitude or activation-based markers in distinguishing cognitive states. These findings reframe task-evoked activation and deactivation as constructive and destructive interference among concurrent flows, rather than selective engagement or disengagement. This reconceptualization implies that the primary control variable for cognition is when intrinsic dynamics align in time, not where or how much the brain activates. Together, these results demonstrate that flexible cognition can emerge from retiming without reconfiguring functional architecture, offering a new path toward understanding the principled link between intrinsic dynamics and task-evoked cognition.

Background: Schizophrenia is a severe neuropsychiatric disorder. Efforts to describe the underlying biology and establish diagnostic markers through non-invasive neuroimaging methods are ongoing, resulting in a range of theoretical brain-based frameworks. Prominent frameworks for aberrant schizophrenia-associated functional connectivity in resting-state functional magnetic resonance imaging (rsfMRI) include the dysconnectivity hypothesis, theory of cognitive dysmetria, and triple network theory. Although informative, prior work can be improved by increasing sample size, avoiding confirmation bias, and accounting for individual variability and the effects of medication and chronicity. Methods: With these recommendations in mind, we employed a data-driven, whole-brain approach using a large multi-site rsfMRI dataset ( N = 2,656; schizophrenia = 1,248). We used reference-guided independent component analysis to generate subject-specific whole-brain functional network connectivity (FNC) and extract imaging markers of similarity to schizophrenia patterns. We modeled the relationship between medication dosage, age of onset, chronicity, symptom severity, and cognitive performance and FNC. Results: Our analysis identified a reliable schizophrenia-FNC signature characterized by aberrantly stronger negative cerebellothalamic and positive thalamocortical connectivity, implicating sensory, motor, and associative cortical circuits. While medication and chronicity were significantly associated with these signatures, the core cerebellothalamic disruptions remained a robust marker of schizophrenia. Conclusions: This work represents the largest schizophrenia-specific rsfMRI study to date, refines existing theoretical frameworks with a more nuanced map of how clinical variables interact with brain connectivity, and provides a high-fidelity template of schizophrenia-related connectivity. We have released this template as an open-access resource to facilitate reproducibility and accelerate the development of reliable rsfMRI-based schizophrenia biomarkers.

Bullying is an adverse childhood experience affecting up to one-third of the global population and linked to psychosis-like experiences (PLEs), which increase the risk of psychotic disorders. This study aimed to investigate the association between the severity and persistence of bullying and PLEs and the neurobiological pathways from bullying to psychosis-like experiences by assessing multiscale brain functional network connectivity (msFNC). We used data from the ABCD Study, a large, ongoing, multisite, population-based prospective cohort study following U.S. adolescents. We included adolescents with complete bullying and PLEs assessments at the 2- and 3-year follow-ups (T1: n=10,939; T2: n=10,102). We examined the associations between bullying severity and temporal exposure and PLEs using linear mixed-effects models. In a 2-year rsfMRI subsample (n=5,280), we used a Neuromark framework to analyze whether msFNC mediated the pathway from bullying to PLEs. Higher PLEs were associated with the presence and severity of bullying (non-bullied vs. mild bullying: d=-0.19, CI: −0.39 to −0.19, p<0.0001; moderate vs. severe bullying: d=-0.49, CI: −0.69 to −0.56, p<0.0001). When bullying ceased, PLEs returned to non-bullied levels (d=-0.13, CI:-0.20 to –0.05, p=0.16), whereas persistence over two years led to greater elevations (d=-0.36, CI:-0.43 to –0.29, p<0.0001). We observed similar patterns for non-paranoid and hallucination-like experiences and their distress. msFNC in paralimbic, default mode, central executive, somatomotor, temporoparietal, insulotemporal, and frontal networks mediated the association. Bullying is time- and dose-dependently associated with psychosis-like outcomes. msFNC between functional brain networks is a novel neurobiological pathway that mediates the link from bullying to PLEs.

High dimensionality of dynamic functional connectivity (dFNC) data representation complicates clinical interpretation and biomarker discovery. We propose a new complementary analytical framework based on dynamic inter-network connectivity entropy (DICE) and its potential use as a biomarker of mental illness. Our framework shows that DICE features extend beyond patient-control discrimination, revealing distinct pathophysiological signatures and differential associations with symptom dimensions. Using resting-state fMRI data from 311 participants, 160 controls, 151 schizophrenia (SZ) patients, we identified 53 intrinsic networks, computed DICE and derived three families of DICE-based metrics: (i) entropy level and range, (ii) distributional shape and temporal organization, (iii) entropy-state repertoire and occupancy. These measures revealed a multidimensional signature of altered entropy dynamics in SZ: (1) elevated baseline entropy with reduced fluctuation magnitude and reduced entropy acceleration; (2) reduced temporal persistence of entropy excursions and entropy distributions closer to Gaussian; and (3) a narrowed repertoire of entropy states, prolonged time in near-baseline entropy configurations. The DICE-based metrics within the SZ group show different associations with symptom dimensions. Reduced fluctuation magnitude and acceleration were associated with greater PANSS general symptom severity (disturbance of volition and preoccupation). Reduced deviation from Gaussianity was associated with higher PANSS positive severity (delusions and hallucinations). Reduced temporal persistence was associated with multiple PANSS positive, negative, and general symptoms. Reduced entropy-state diversity and prolonged dwell time in near-baseline states were associated with depression and PANSS positive/general severity, respectively. The multidimensional pathophysiology revealed through the different entropy patterns may potentially guide biomarker development and personalized treatments.