Psychosis brain subtypes validated in first-episode cohorts and related to illness remission: results from the PHENOM consortium.

Using machine learning, we recently decomposed the neuroanatomical heterogeneity of established schizophrenia to discover two volumetric subgroups-a 'lower brain volume' subgroup (SG1) and an 'higher striatal volume' subgroup (SG2) with otherwise normal brain structure. In this study, we investigated whether the MRI signatures of these subgroups were also already present at the time of the first-episode of psychosis (FEP) and whether they were related to clinical presentation and clinical remission over 1-, 3-, and 5-years. We included 572 FEP and 424 healthy controls (HC) from 4 sites (Sao Paulo, Santander, London, Melbourne) of the PHENOM consortium. Our prior MRI subgrouping models (671 participants; USA, Germany, and China) were applied to both FEP and HC. Participants were assigned into 1 of 4 categories: subgroup 1 (SG1), subgroup 2 (SG2), no subgroup membership ('None'), and mixed SG1 + SG2 subgroups ('Mixed'). Voxel-wise analyses characterized SG1 and SG2 subgroups. Supervised machine learning analyses characterized baseline and remission signatures related to SG1 and SG2 membership. The two dominant patterns of 'lower brain volume' in SG1 and 'higher striatal volume' (with otherwise normal neuromorphology) in SG2 were identified already at the first episode of psychosis. SG1 had a significantly higher proportion of FEP (32%) vs. HC (19%) than SG2 (FEP, 21%; HC, 23%). Clinical multivariate signatures separated the SG1 and SG2 subgroups (balanced accuracy = 64%; p < 0.0001), with SG2 showing higher education but also greater positive psychosis symptoms at first presentation, and an association with symptom remission at 1-year, 5-year, and when timepoints were combined. Neuromorphological subtypes of schizophrenia are already evident at illness onset, separated by distinct clinical presentations, and differentially associated with subsequent remission. These results suggest that the subgroups may be underlying risk phenotypes that could be targeted in future treatment trials and are critical to consider when interpreting neuroimaging literature.

Two distinct neuroanatomical subtypes of schizophrenia revealed using machine learning.

Neurobiological heterogeneity in schizophrenia is poorly understood and confounds current analyses. We investigated neuroanatomical subtypes in a multi-institutional multi-ethnic cohort, using novel semi-supervised machine learning methods designed to discover patterns associated with disease rather than normal anatomical variation. Structural MRI and clinical measures in established schizophrenia (n = 307) and healthy controls (n = 364) were analysed across three sites of PHENOM (Psychosis Heterogeneity Evaluated via Dimensional Neuroimaging) consortium. Regional volumetric measures of grey matter, white matter, and CSF were used to identify distinct and reproducible neuroanatomical subtypes of schizophrenia. Two distinct neuroanatomical subtypes were found. Subtype 1 showed widespread lower grey matter volumes, most prominent in thalamus, nucleus accumbens, medial temporal, medial prefrontal/frontal and insular cortices. Subtype 2 showed increased volume in the basal ganglia and internal capsule, and otherwise normal brain volumes. Grey matter volume correlated negatively with illness duration in Subtype 1 (r = -0.201, P = 0.016) but not in Subtype 2 (r = -0.045, P = 0.652), potentially indicating different underlying neuropathological processes. The subtypes did not differ in age (t = -1.603, df = 305, P = 0.109), sex (chi-square = 0.013, df = 1, P = 0.910), illness duration (t = -0.167, df = 277, P = 0.868), antipsychotic dose (t = -0.439, df = 210, P = 0.521), age of illness onset (t = -1.355, df = 277, P = 0.177), positive symptoms (t = 0.249, df = 289, P = 0.803), negative symptoms (t = 0.151, df = 289, P = 0.879), or antipsychotic type (chi-square = 6.670, df = 3, P = 0.083). Subtype 1 had lower educational attainment than Subtype 2 (chi-square = 6.389, df = 2, P = 0.041). In conclusion, we discovered two distinct and highly reproducible neuroanatomical subtypes. Subtype 1 displayed widespread volume reduction correlating with illness duration, and worse premorbid functioning. Subtype 2 had normal and stable anatomy, except for larger basal ganglia and internal capsule, not explained by antipsychotic dose. These subtypes challenge the notion that brain volume loss is a general feature of schizophrenia and suggest differential aetiologies. They can facilitate strategies for clinical trial enrichment and stratification, and precision diagnostics.

Disrupted interactions among the hippocampal, dorsal attention, and central-executive networks in amnestic mild cognitive impairment.

Neuroimaging investigations consistently demonstrate that the neural processes involve complex interactions between the large-scale networks. Among those networks, the dorsal attention network (DAN) and the central-executive network (CEN) have been previously shown to exhibit anti-correlated activity with the default-mode network (DMN) in cognitively normal people. In amnestic mild cognitive impairment (MCI) and Alzheimer's disease, the hippocampal network (HCN)-a key memory processing system-and its interactions with other networks have gathered central interest. The current study aims to evaluate the patterns of functional architectures of the HCN with the three networks-DAN, CEN, and DMN-in amnestic MCI and normal controls (NC) to test the hypothesis that the interactions of HCN with other networks alter in MCI. We recorded the resting state functional MRI, assessed patterns of functional architectures between the four networks using dynamical causal modeling, and compared between NC and MCI. Our analysis showed that the DAN modulates the activity between the HCN and the DMN in both MCI and NC. We further uncovered that the DAN modulates the activity between the HCN and the CEN in NC, but such modulation is impaired in MCI. We found an association between impaired modulation and Montreal cognitive assessment (R = 0.349). Overall, our findings provide important insight in understanding the neuroimaging signature of amnestic MCI and/or Alzheimer's disease.

Interactions between the anterior cingulate-insula network and the fronto-parietal network during perceptual decision-making.

Information processing in the human brain during cognitively demanding goal-directed tasks is thought to involve several large-scale brain networks, including the anterior cingulate-insula network (aCIN) and the fronto-parietal network (FPN). Recent functional MRI (fMRI) studies have provided clues that the aCIN initiates activity changes in the FPN. However, when and how often these networks interact remains largely unknown to date. Here, we systematically examined the oscillatory interactions between the aCIN and the FPN by using the spectral Granger causality analysis of reconstructed brain source signals from the scalp electroencephalography (EEG) recorded from human participants performing a face-house perceptual categorization task. We investigated how the aCIN and the FPN interact, what the temporal sequence of events in these nodes is, and what frequency bands of information flow bind these nodes in networks. We found that beta band (13-30Hz) and gamma (30-100Hz) bands of interactions are involved between the aCIN and the FPN during decision-making tasks. In gamma band, the aCIN initiated the Granger causal control over the FPN in 25-225 ms timeframe. In beta band, the FPN achieved a control over the aCIN in 225-425 ms timeframe. These band-specific time-dependent Granger causal controls of the aCIN and the FPN were retained for behaviorally harder decision-making tasks. These findings of times and frequencies of oscillatory interactions in the aCIN and FPN provide us new insights into the general neural mechanisms for sensory information-guided, goal-directed behaviors, including perceptual decision-making processes.

The salience network dynamics in perceptual decision-making.

Recent neuroimaging studies have demonstrated that the network consisting of the right anterior insula (rAI), left anterior insula (lAI) and dorsal anterior cingulate cortex (dACC) is activated in sensory stimulus-guided goal-directed behaviors. This network is often known as the salience network (SN). When and how a sensory signal enters and organizes within SN before reaching the central executive network including the prefrontal cortices is still a mystery. Previous electrophysiological studies focused on individual nodes of SN, either on dACC or rAI, have reports of conflicting findings of the earliest cortical activity within the network. Functional magnetic resonance imaging (fMRI) studies are not able to answer these questions in the time-scales of human sensory perception and decision-making. Here, using clear and noisy face-house image categorization tasks and human scalp electroencephalography (EEG) recordings combined with source reconstruction techniques, we study when and how oscillatory activity organizes SN during a perceptual decision. We uncovered that the beta-band (13-30Hz) oscillations bound SN, became most active around 100ms after the stimulus onset and the rAI acted as a main outflow hub within SN for easier decision making task. The SN activities (Granger causality measures) were negatively correlated with the decision response time (decision difficulty). These findings suggest that the SN activity precedes the executive control in mediating sensory and cognitive processing to arrive at visual perceptual decisions.

Dysregulated Salience Network Control over Default-Mode and Central-Executive Networks in Schizophrenia Revealed Using Stochastic Dynamical Causal Modeling.

Introduction: Neuroimaging studies suggest that the human brain consists of intrinsically organized, large-scale neural networks. Among these networks, the interplay among the default-mode network (DMN), salience network (SN), and central-executive network (CEN) has been widely used to understand the functional interaction patterns in health and disease. This triple network model suggests that the SN causally controls over the DMN and CEN in healthy individuals. This interaction is often referred to as SN's dynamic regulating mechanism. However, such interactions are not well understood in individuals with schizophrenia. Methods: In this study, we leveraged resting-state functional magnetic resonance imaging data from schizophrenia (n = 67) and healthy controls (n = 81) and evaluated the directional functional interactions among DMN, SN, and CEN using stochastic dynamical causal modeling methodology. Results: In healthy controls, our analyses replicated previous findings that SN regulates DMN and CEN activities (Mann-Whitney U test; p < 10-8). In schizophrenia, however, our analyses revealed a disrupted SN-based controlling mechanism over the DMN and CEN (Mann-Whitney U test; p < 10-16). Conclusions: These results indicate that the disrupted controlling mechanism of SN over the other two neural networks may be a candidate neuroimaging phenotype in schizophrenia.

Explainable AI-based Deep-SHAP for mapping the multivariate relationships between regional neuroimaging biomarkers and cognition.

BACKGROUND: Mild cognitive impairment (MCI)/Alzheimer's disease (AD) is associated with cognitive decline beyond normal aging and linked to the alterations of brain volume quantified by magnetic resonance imaging (MRI) and amyloid-beta (Aß) quantified by positron emission tomography (PET). Yet, the complex relationships between these regional imaging measures and cognition in MCI/AD remain unclear. Explainable artificial intelligence (AI) may uncover such relationships. METHOD: We integrate the AI-based deep learning neural network and Shapley additive explanations (SHAP) approaches and introduce the Deep-SHAP method to investigate the multivariate relationships between regional imaging measures and cognition. After validating this approach on simulated data, we apply it to real experimental data from MCI/AD patients. RESULTS: Deep-SHAP significantly predicted cognition using simulated regional features and identified the ground-truth simulated regions as the most significant multivariate predictors. When applied to experimental MRI data, Deep-SHAP revealed that the insula, lateral occipital, medial frontal, temporal pole, and occipital fusiform gyrus are the primary contributors to global cognitive decline in MCI/AD. Furthermore, when applied to experimental amyloid Pittsburgh compound B (PiB)-PET data, Deep-SHAP identified the key brain regions for global cognitive decline in MCI/AD as the inferior temporal, parahippocampal, inferior frontal, supratemporal, and lateral frontal gray matter. CONCLUSION: Deep-SHAP method uncovered the multivariate relationships between regional brain features and cognition, offering insights into the most critical modality-specific brain regions involved in MCI/AD mechanisms.

Predicting cognitive dysfunction and regional hubs using Braak staging amyloid-beta biomarkers and machine learning.

Mild cognitive impairment (MCI) is a transitional stage between normal aging and early Alzheimer's disease (AD). The presence of extracellular amyloid-beta (Aß) in Braak regions suggests a connection with cognitive dysfunction in MCI/AD. Investigating the multivariate predictive relationships between regional Aß biomarkers and cognitive function can aid in the early detection and prevention of AD. We introduced machine learning approaches to estimate cognitive dysfunction from regional Aß biomarkers and identify the Aß-related dominant brain regions involved with cognitive impairment. We employed Aß biomarkers and cognitive measurements from the same individuals to train support vector regression (SVR) and artificial neural network (ANN) models and predict cognitive performance solely based on Aß biomarkers on the test set. To identify Aß-related dominant brain regions involved in cognitive prediction, we built the local interpretable model-agnostic explanations (LIME) model. We found elevated Aß in MCI compared to controls and a stronger correlation between Aß and cognition, particularly in Braak stages III-IV and V-VII (p < 0.05) biomarkers. Both SVR and ANN, especially ANN, showed strong predictive relationships between regional Aß biomarkers and cognitive impairment (p < 0.05). LIME integrated with ANN showed that the parahippocampal gyrus, inferior temporal gyrus, and hippocampus were the most decisive Braak regions for predicting cognitive decline. Consistent with previous findings, this new approach suggests relationships between Aß biomarkers and cognitive impairment. The proposed analytical framework can estimate cognitive impairment from Braak staging Aß biomarkers and delineate the dominant brain regions collectively involved in AD pathophysiology.

Two Neuroanatomical Signatures in Schizophrenia: Expression Strengths Over the First 2 Years of Treatment and Their Relationships to Neurodevelopmental Compromise and Antipsychotic Treatment.

BACKGROUND AND HYPOTHESIS: Two machine learning derived neuroanatomical signatures were recently described. Signature 1 is associated with widespread grey matter volume reductions and signature 2 with larger basal ganglia and internal capsule volumes. We hypothesized that they represent the neurodevelopmental and treatment-responsive components of schizophrenia respectively. STUDY DESIGN: We assessed the expression strength trajectories of these signatures and evaluated their relationships with indicators of neurodevelopmental compromise and with antipsychotic treatment effects in 83 previously minimally treated individuals with a first episode of a schizophrenia spectrum disorder who received standardized treatment and underwent comprehensive clinical, cognitive and neuroimaging assessments over 24 months. Ninety-six matched healthy case-controls were included. STUDY RESULTS: Linear mixed effect repeated measures models indicated that the patients had stronger expression of signature 1 than controls that remained stable over time and was not related to treatment. Stronger signature 1 expression showed trend associations with lower educational attainment, poorer sensory integration, and worse cognitive performance for working memory, verbal learning and reasoning and problem solving. The most striking finding was that signature 2 expression was similar for patients and controls at baseline but increased significantly with treatment in the patients. Greater increase in signature 2 expression was associated with larger reductions in PANSS total score and increases in BMI and not associated with neurodevelopmental indices. CONCLUSIONS: These findings provide supporting evidence for two distinct neuroanatomical signatures representing the neurodevelopmental and treatment-responsive components of schizophrenia.

Neuroimaging-AI Endophenotypes of Brain Diseases in the General Population: Towards a Dimensional System of Vulnerability.

Disease heterogeneity poses a significant challenge for precision diagnostics in both clinical and sub-clinical stages. Recent work leveraging artificial intelligence (AI) has offered promise to dissect this heterogeneity by identifying complex intermediate phenotypes - herein called dimensional neuroimaging endophenotypes (DNEs) - which subtype various neurologic and neuropsychiatric diseases. We investigate the presence of nine such DNEs derived from independent yet harmonized studies on Alzheimer's disease (AD1-2)1, autism spectrum disorder (ASD1-3)2, late-life depression (LLD1-2)3, and schizophrenia (SCZ1-2)4, in the general population of 39,178 participants in the UK Biobank study. Phenome-wide associations revealed prominent associations between the nine DNEs and phenotypes related to the brain and other human organ systems. This phenotypic landscape aligns with the SNP-phenotype genome-wide associations, revealing 31 genomic loci associated with the nine DNEs (Bonferroni corrected P-value < 5×10-8/9). The DNEs exhibited significant genetic correlations, colocalization, and causal relationships with multiple human organ systems and chronic diseases. A causal effect (odds ratio=1.25 [1.11, 1.40], P-value=8.72×1-4) was established from AD2, characterized by focal medial temporal lobe atrophy, to AD. The nine DNEs and their polygenic risk scores significantly improved the prediction accuracy for 14 systemic disease categories and mortality. These findings underscore the potential of the nine DNEs to identify individuals at a high risk of developing the four brain diseases during preclinical stages for precision diagnostics. All results are publicly available at: http://labs.loni.usc.edu/medicine/.

Assessment of Neuroanatomical Endophenotypes of Autism Spectrum Disorder and Association With Characteristics of Individuals With Schizophrenia and the General Population.

Importance: Autism spectrum disorder (ASD) is associated with significant clinical, neuroanatomical, and genetic heterogeneity that limits precision diagnostics and treatment. Objective: To assess distinct neuroanatomical dimensions of ASD using novel semisupervised machine learning methods and to test whether the dimensions can serve as endophenotypes also in non-ASD populations. Design, Setting, and Participants: This cross-sectional study used imaging data from the publicly available Autism Brain Imaging Data Exchange (ABIDE) repositories as the discovery cohort. The ABIDE sample included individuals diagnosed with ASD aged between 16 and 64 years and age- and sex-match typically developing individuals. Validation cohorts included individuals with schizophrenia from the Psychosis Heterogeneity Evaluated via Dimensional Neuroimaging (PHENOM) consortium and individuals from the UK Biobank to represent the general population. The multisite discovery cohort included 16 internationally distributed imaging sites. Analyses were performed between March 2021 and March 2022. Main Outcomes and Measures: The trained semisupervised heterogeneity through discriminative analysis models were tested for reproducibility using extensive cross-validations. It was then applied to individuals from the PHENOM and the UK Biobank. It was hypothesized that neuroanatomical dimensions of ASD would display distinct clinical and genetic profiles and would be prominent also in non-ASD populations. Results: Heterogeneity through discriminative analysis models trained on T1-weighted brain magnetic resonance images of 307 individuals with ASD (mean [SD] age, 25.4 [9.8] years; 273 [88.9%] male) and 362 typically developing control individuals (mean [SD] age, 25.8 [8.9] years; 309 [85.4%] male) revealed that a 3-dimensional scheme was optimal to capture the ASD neuroanatomy. The first dimension (A1: aginglike) was associated with smaller brain volume, lower cognitive function, and aging-related genetic variants (FOXO3; Z = 4.65; P = 1.62 × 10-6). The second dimension (A2: schizophrenialike) was characterized by enlarged subcortical volumes, antipsychotic medication use (Cohen d = 0.65; false discovery rate-adjusted P = .048), partially overlapping genetic, neuroanatomical characteristics to schizophrenia (n = 307), and significant genetic heritability estimates in the general population (n = 14 786; mean [SD] h2, 0.71 [0.04]; P < 1 × 10-4). The third dimension (A3: typical ASD) was distinguished by enlarged cortical volumes, high nonverbal cognitive performance, and biological pathways implicating brain development and abnormal apoptosis (mean [SD] ß, 0.83 [0.02]; P = 4.22 × 10-6). Conclusions and Relevance: This cross-sectional study discovered 3-dimensional endophenotypic representation that may elucidate the heterogeneous neurobiological underpinnings of ASD to support precision diagnostics. The significant correspondence between A2 and schizophrenia indicates a possibility of identifying common biological mechanisms across the 2 mental health diagnoses.

Salience network anatomical and molecular markers are linked with cognitive dysfunction in mild cognitive impairment.

BACKGROUND AND PURPOSE: Recent studies indicate disrupted functional mechanisms of salience network (SN) regions-right anterior insula, left anterior insula, and anterior cingulate cortex-in mild cognitive impairment (MCI). However, the underlying anatomical and molecular mechanisms in these regions are not clearly understood yet. It is also unknown whether integration of multimodal-anatomical and molecular-markers could predict cognitive impairment better in MCI. METHODS: Herein we quantified anatomical volumetric markers via structural MRI and molecular amyloid markers via PET with Pittsburgh compound B in SN regions of MCI (n = 33) and healthy controls (n = 27). From these markers, we built support vector machine learning models aiming to estimate cognitive dysfunction in MCI. RESULTS: We found that anatomical markers are significantly reduced and molecular markers are significantly elevated in SN nodes of MCI compared to healthy controls (p < .05). These altered markers in MCI patients were associated with their worse cognitive performance (p < .05). Our machine learning-based modeling further suggested that the integration of multimodal markers predicts cognitive impairment in MCI superiorly compared to using single modality-specific markers. CONCLUSIONS: These findings shed light on the underlying anatomical volumetric and molecular amyloid alterations in SN regions and show the significance of multimodal markers integration approach in better predicting cognitive impairment in MCI.

Differential Sphingosine-1-Phosphate Receptor-1 Protein Expression in the Dorsolateral Prefrontal Cortex Between Schizophrenia Type 1 and Type 2.

Understanding the etiology and treatment approaches in schizophrenia is challenged in part by the heterogeneity of this disorder. One encouraging progress is the growing evidence that there are subtypes of schizophrenia. Recent in vitro findings of messenger ribonucleic acid (mRNA) gene expression on postmortem dorsolateral prefrontal cortex (DLPFC) showed that schizophrenia has two subtypes, those with a relatively normal DLPFC transcriptome (Type 1) and those with differentially expressed genes (Type 2). Sphingosine-1-phosphate receptor-1 (S1PR1) is one of the genes that was highly upregulated in Type 2 compared to Type 1 and controls. The impact of that finding is limited because it only can be confirmed through analysis of autopsy tissue, and the clinical characteristics such as symptoms severity or illness duration except for cause of death was not available from that Medical Examiner based autopsy study. However, S1PR1 has great potential because it is a target gene that can be accessed via positron emission tomography (PET) in vivo using specific radioligands (starting with [11C]CS1P1) successfully developed at our center in human brain imaging. As a preliminary study to validate this PET target in schizophrenia, S1PR1 protein expression was assessed by receptor autoradiography (ARG) using [3H]CS1P1 and immunohistochemistry (IHC) in the DLPFC from patients with schizophrenia classified as Type 1 or Type 2 based on their DLPFC transcriptomes and from controls. Our analyses demonstrate that ARG S1PR1 protein expression is significantly higher in Type 2 compared to Type 1 (p < 0.05) and controls (p < 0.05), which was consistent with previous mRNA S1PR1. These findings support the possibility that PET S1PR1 can be used as a future imaging biomarker to distinguish these subgroups of schizophrenic patients during life with obvious implications for both patient management and the design of clinical trials to validate novel pharmacologic therapies.

Schizophrenia Imaging Signatures and Their Associations With Cognition, Psychopathology, and Genetics in the General Population.

OBJECTIVE: The prevalence and significance of schizophrenia-related phenotypes at the population level is debated in the literature. Here, the authors assessed whether two recently reported neuroanatomical signatures of schizophrenia-signature 1, with widespread reduction of gray matter volume, and signature 2, with increased striatal volume-could be replicated in an independent schizophrenia sample, and investigated whether expression of these signatures can be detected at the population level and how they relate to cognition, psychosis spectrum symptoms, and schizophrenia genetic risk. METHODS: This cross-sectional study used an independent schizophrenia-control sample (N=347; ages 16-57 years) for replication of imaging signatures, and then examined two independent population-level data sets: typically developing youths and youths with psychosis spectrum symptoms in the Philadelphia Neurodevelopmental Cohort (N=359; ages 16-23 years) and adults in the UK Biobank study (N=836; ages 44-50 years). The authors quantified signature expression using support-vector machine learning and compared cognition, psychopathology, and polygenic risk between signatures. RESULTS: Two neuroanatomical signatures of schizophrenia were replicated. Signature 1 but not signature 2 was significantly more common in youths with psychosis spectrum symptoms than in typically developing youths, whereas signature 2 frequency was similar in the two groups. In both youths and adults, signature 1 was associated with worse cognitive performance than signature 2. Compared with adults with neither signature, adults expressing signature 1 had elevated schizophrenia polygenic risk scores, but this was not seen for signature 2. CONCLUSIONS: The authors successfully replicated two neuroanatomical signatures of schizophrenia and describe their prevalence in population-based samples of youths and adults. They further demonstrated distinct relationships of these signatures with psychosis symptoms, cognition, and genetic risk, potentially reflecting underlying neurobiological vulnerability.

Multi-scale semi-supervised clustering of brain images: Deriving disease subtypes.

Disease heterogeneity is a significant obstacle to understanding pathological processes and delivering precision diagnostics and treatment. Clustering methods have gained popularity for stratifying patients into subpopulations (i.e., subtypes) of brain diseases using imaging data. However, unsupervised clustering approaches are often confounded by anatomical and functional variations not related to a disease or pathology of interest. Semi-supervised clustering techniques have been proposed to overcome this and, therefore, capture disease-specific patterns more effectively. An additional limitation of both unsupervised and semi-supervised conventional machine learning methods is that they typically model, learn and infer from data using a basis of feature sets pre-defined at a fixed anatomical or functional scale (e.g., atlas-based regions of interest). Herein we propose a novel method, "Multi-scAle heteroGeneity analysIs and Clustering" (MAGIC), to depict the multi-scale presentation of disease heterogeneity, which builds on a previously proposed semi-supervised clustering method, HYDRA. It derives multi-scale and clinically interpretable feature representations and exploits a double-cyclic optimization procedure to effectively drive identification of inter-scale-consistent disease subtypes. More importantly, to understand the conditions under which the clustering model can estimate true heterogeneity related to diseases, we conducted extensive and systematic semi-simulated experiments to evaluate the proposed method on a sizeable healthy control sample from the UK Biobank (N = 4403). We then applied MAGIC to imaging data from Alzheimer's disease (ADNI, N = 1728) and schizophrenia (PHENOM, N = 1166) patients to demonstrate its potential and challenges in dissecting the neuroanatomical heterogeneity of common brain diseases. Taken together, we aim to provide guidance regarding when such analyses can succeed or should be taken with caution. The code of the proposed method is publicly available at https://github.com/anbai106/MAGIC.

Neurocognitive and functional heterogeneity in depressed youth.

Depression is a common psychiatric illness that often begins in youth, and is sometimes associated with cognitive deficits. However, there is significant variability in cognitive dysfunction, likely reflecting biological heterogeneity. We sought to identify neurocognitive subtypes and their neurofunctional signatures in a large cross-sectional sample of depressed youth. Participants were drawn from the Philadelphia Neurodevelopmental Cohort, including 712 youth with a lifetime history of a major depressive episode and 712 typically developing (TD) youth matched on age and sex. A subset (MDD n = 368, TD n = 200) also completed neuroimaging. Cognition was assessed with the Penn Computerized Neurocognitive Battery. A recently developed semi-supervised machine learning algorithm was used to delineate neurocognitive subtypes. Subtypes were evaluated for differences in both clinical psychopathology and brain activation during an n-back working memory fMRI task. We identified three neurocognitive subtypes in the depressed group. Subtype 1 was high-performing (high accuracy, moderate speed), Subtype 2 was cognitively impaired (low accuracy, slow speed), and Subtype 3 was impulsive (low accuracy, fast speed). While subtypes did not differ in clinical psychopathology, they diverged in their activation profiles in regions critical for executive function, which mirrored differences in cognition. Taken together, these data suggest disparate mechanisms of cognitive vulnerability and resilience in depressed youth, which may inform the identification of biomarkers for prognosis and treatment response.

Estimating regional cerebral blood flow using resting-state functional MRI via machine learning.

BACKGROUND: Perfusion MRI is an important modality in many brain imaging protocols, since it probes cerebrovascular changes in aging and many diseases; however, it may not be always available. NEW METHOD: We introduce a new method that seeks to estimate regional perfusion properties using spectral information of resting-state functional MRI (rsfMRI) via machine learning. We used pairs of rsfMRI and arterial spin labeling (ASL) images from the same individuals with normal cognition and mild cognitive impairment (MCI), and built support vector machine models aiming to estimate regional cerebral blood flow (CBF) from the rsfMRI signal alone. RESULTS: This method demonstrated higher associations between the estimated CBF and actual CBF (ASL-CBF) at the total lobar gray matter (r = 0.40; FDR-p = 1.9e-03), parietal lobe (r = 0.46, FDR-p = 8e-04), and occipital lobe (r = 0.35; FDR-p = 0.01) using rsfMRI signals of frequencies [0.01-0.15] Hertz compared to frequencies [0.01-0.10] Hertz and [0.01-0.20] Hertz. We further observed significant associations between the estimated CBF and actual CBF in 24 regions of interest (p < 0.05), with the highest association observed in the superior parietal lobule (r = 0.50, FDR-p = 0.002). Moreover, the estimated CBF at superior parietal lobule showed significant correlation with the mini-mental state exam (MMSE) score (r = 0.27; FDR-p = 0.04) and decreased in MCI with lower MMSE score compared to NC group (FDR-p = 0.04). COMPARISON WITH EXISTING METHODS: Consistent with previous findings, this new method also suggests that rsfMRI signals contain perfusion information. CONCLUSION: The proposed framework can obtain estimates of regional perfusion from rsfMRI, which can serve as surrogate perfusion measures in the absence of ASL.

Neurostructural Heterogeneity in Youths With Internalizing Symptoms.

BACKGROUND: Internalizing disorders such as anxiety and depression are common psychiatric disorders that frequently begin in youth and exhibit marked heterogeneity in treatment response and clinical course. Given that symptom-based classification approaches do not align with underlying neurobiology, an alternative approach is to identify neurobiologically informed subtypes based on brain imaging data. METHODS: We used a recently developed semisupervised machine learning method (HYDRA [heterogeneity through discriminative analysis]) to delineate patterns of neurobiological heterogeneity within youths with internalizing symptoms using structural data collected at 3T from a sample of 1141 youths. RESULTS: Using volume and cortical thickness, cross-validation methods indicated 2 highly stable subtypes of internalizing youths (adjusted Rand index = 0.66; permutation-based false discovery rate p < .001). Subtype 1, defined by smaller brain volumes and reduced cortical thickness, was marked by impaired cognitive performance and higher levels of psychopathology than both subtype 2 and typically developing youths. Using resting-state functional magnetic resonance imaging and diffusion images not considered during clustering, we found that subtype 1 also showed reduced amplitudes of low-frequency fluctuations in frontolimbic regions at rest and reduced fractional anisotropy in several white matter tracts. In contrast, subtype 2 showed intact cognitive performance and greater volume, cortical thickness, and amplitudes during rest compared with subtype 1 and typically developing youths, despite still showing clinically significant levels of psychopathology. CONCLUSIONS: We identified 2 subtypes of internalizing youths differentiated by abnormalities in brain structure, function, and white matter integrity, with one of the subtypes showing poorer functioning across multiple domains. Identification of biologically grounded internalizing subtypes may assist in targeting early interventions and assessing longitudinal prognosis.

Interactions of Insula Subdivisions-Based Networks with Default-Mode and Central-Executive Networks in Mild Cognitive Impairment.

Interactions between the brain networks and subnetworks are crucial for active and resting cognitive states. Whether a subnetwork can restore the adequate function of the parent network whenever a disease state affects the parent network is unclear. Investigations suggest that the control of the anterior insula-based network (AIN) over the default-mode network (DMN) and central-executive network (CEN) is decreased in individuals with mild cognitive impairment (MCI). Here, we hypothesized that the posterior insula-based network (PIN) attempts to compensate for this decrease. To test this, we compared a group of MCI and normal cognitive individuals. A dynamical causal modeling method has been employed to investigate the dynamic network controls/modulations. We used the resting state functional MRI data, and assessed the interactions of the AIN and of the PIN, respectively, over the DMN and CEN. We found that the greater control of AIN than that of DMN (Wilcoxon rank sum: Z = 1.987; p = 0.047) and CEN (Z = 3.076; p = 0.002) in normal group and the lower (impaired) control of AIN than that of CEN (Z = 8.602; p = 7.816 × 10-18). We further revealed that the PIN control was significantly higher than that of DMN (Z = 6.608; p = 3.888 × 10-11) and CEN (Z = 6.429; p = 1.278 × 10-10) in MCI group where the AIN was impaired, but that control was significantly lower than of DMN (Z = 5.285; p = 1.254 × 10-7) and CEN (Z = 5.404; p = 6.513 × 10-8) in normal group. Finally, the global cognitive test score assessed using Montreal cognitive assessment and the network modulations were correlated (Spearman's correlation: r = 0.47; p = 3.76 × 10-5 and r = -0.43; p = 1.97 × 10-4). These findings might suggest the flexible functional profiles of AIN and PIN in normal aging and MCI.

Interactions of the Salience Network and Its Subsystems with the Default-Mode and the Central-Executive Networks in Normal Aging and Mild Cognitive Impairment.

Previous functional magnetic resonance imaging (fMRI) investigations suggest that the intrinsically organized large-scale networks and the interaction between them might be crucial for cognitive activities. A triple network model, which consists of the default-mode network, salience network, and central-executive network, has been recently used to understand the connectivity patterns of the cognitively normal brains versus the brains with disorders. This model suggests that the salience network dynamically controls the default-mode and central-executive networks in healthy young individuals. However, the patterns of interactions have remained largely unknown in healthy aging or those with cognitive decline. In this study, we assess the patterns of interactions between the three networks using dynamical causal modeling in resting state fMRI data and compare them between subjects with normal cognition and mild cognitive impairment (MCI). In healthy elderly subjects, our analysis showed that the salience network, especially its dorsal subnetwork, modulates the interaction between the default-mode network and the central-executive network (Mann-Whitney U test; p < 0.05), which was consistent with the pattern of interaction reported in young adults. In contrast, this pattern of modulation by salience network was disrupted in MCI (p < 0.05). Furthermore, the degree of disruption in salience network control correlated significantly with lower overall cognitive performance measured by Montreal Cognitive Assessment (r = 0.295; p < 0.05). This study suggests that a disruption of the salience network control, especially the dorsal salience network, over other networks provides a neuronal basis for cognitive decline and may be a candidate neuroimaging biomarker of cognitive impairment.