Multivariate classification of blood oxygen level-dependent FMRI data with diagnostic intention: a clinical perspective

Seeing more than the Tip of the Iceberg: Approaches to Subthreshold Effects in Functional Magnetic Resonance Imaging of the Brain

Many functional magnetic resonance imaging (fMRI) studies and presurgical mapping applications rely on mass-univariate inference with subsequent multiple comparison correction. Statistical results are frequently visualized as thresholded statistical maps. This approach has inherent limitations including the risk of drawing overly-selective conclusions based only on selective results passing such thresholds. This article gives an overview of both established and newly emerging scientific approaches to supplement such conventional analyses by incorporating information about subthreshold effects with the aim to improve interpretation of findings or leverage a wider array of information. Topics covered include neuroimaging data visualization, p-value histogram analysis and the related Higher Criticism approach for detecting rare and weak effects. Further examples from multivariate analyses and dedicated Bayesian approaches are provided.

Emotion regulation in bipolar disorder type-I: multivariate analysis of fMRI data

BACKGROUND

Bipolar disorder type-I (BD-I) patients are known to show emotion regulation abnormalities. In a previous fMRI study using an explicit emotion regulation paradigm, we compared responses from 19 BD-I patients and 17 matched healthy controls (HC). A standard general linear model-based univariate analysis revealed that BD patients showed increased activations in inferior frontal gyrus when instructed to decrease their emotional response as elicited by neutral images. We implemented multivariate pattern recognition analyses on the same data to examine if we could classify conditions within-group as well as HC versus BD.

METHODS

We reanalyzed explicit emotion regulation data using a multivariate pattern recognition approach, as implemented in PRONTO software. The original experimental paradigm consisted of a full 2 × 2 factorial design, with valence (Negative/Neutral) and instruction (Look/Decrease) as within subject factors.

RESULTS

The multivariate models were able to accurately classify different task conditions when HC and BD were analyzed separately (63.24%-75.00%, p = 0.001-0.012). In addition, the models were able to correctly classify HC versus BD with significant accuracy in conditions where subjects were instructed to downregulate their felt emotion (59.60%-60.84%, p = 0.014-0.018). The results for HC versus BD classification demonstrated contributions from the salience network, several occipital and frontal regions, inferior parietal lobes, as well as other cortical regions, to achieve above-chance classifications.

CONCLUSIONS

Our multivariate analysis successfully reproduced some of the main results obtained in the previous univariate analysis, confirming that these findings are not dependent on the analysis approach. In particular, both types of analyses suggest that there is a significant difference of neural patterns between conditions within each subject group. The multivariate approach also revealed that reappraisal conditions provide the most informative activity for differentiating HC versus BD, irrespective of emotional valence (negative or neutral). The current results illustrate the importance of investigating the cognitive control of emotion in BD. We also propose a set of candidate regions for further study of emotional control in BD.

Brain correlates of subjective cognitive complaints in COVID-19 survivors: A multimodal magnetic resonance imaging study

Cognitive impairment represents a leading residual symptom of COVID-19 infection, which lasts for months after the virus clearance. Up-to-date scientific reports documented a wide spectrum of brain changes in COVID-19 survivors following the illness's resolution, mainly related to neurological and neuropsychiatric consequences. Preliminary insights suggest abnormal brain metabolism, microstructure, and functionality as neural under-layer of post-acute cognitive dysfunction. While previous works focused on brain correlates of impaired cognition as objectively assessed, herein we investigated long-term neural correlates of subjective cognitive decline in a sample of 58 COVID-19 survivors with a multimodal imaging approach. Diffusion Tensor Imaging (DTI) analyses revealed widespread white matter disruption in the sub-group of cognitive complainers compared to the non-complainer one, as indexed by increased axial, radial, and mean diffusivity in several commissural, projection and associative fibres. Likewise, the Multivoxel Pattern Connectivity analysis (MVPA) revealed highly discriminant patterns of functional connectivity in resting-state among the two groups in the right frontal pole and in the middle temporal gyrus, suggestive of inefficient dynamic modulation of frontal brain activity and possible metacognitive dysfunction at rest. Beyond COVID-19 actual pathophysiological brain processes, our findings point toward brain connectome disruption conceivably translating into clinical post-COVID cognitive symptomatology. Our results could pave the way for a potential brain signature of cognitive complaints experienced by COVID-19 survivors, possibly leading to identify early therapeutic targets and thus mitigating its detrimental long-term impact on quality of life in the post-COVID-19 stages.

Distinct Neural Profiles of Frontoparietal Networks in Boys with ADHD and Boys with Persistent Depressive Disorder

Working memory deficits are common in attention-deficit/hyperactivity disorder (ADHD) and depression-two common neurodevelopmental disorders with overlapping cognitive profiles but distinct clinical presentation. Multivariate techniques have previously been utilized to understand working memory processes in functional brain networks in healthy adults but have not yet been applied to investigate how working memory processes within the same networks differ within typical and atypical developing populations. We used multivariate pattern analysis (MVPA) to identify whether brain networks discriminated between spatial versus verbal working memory processes in ADHD and Persistent Depressive Disorder (PDD). Thirty-six male clinical participants and 19 typically developing (TD) boys participated in a fMRI scan while completing a verbal and a spatial working memory task. Within a priori functional brain networks (frontoparietal, default mode, salience), the TD group demonstrated differential response patterns to verbal and spatial working memory. The PDD group showed weaker differentiation than TD, with lower classification accuracies observed in primarily the left frontoparietal network. The neural profiles of the ADHD and PDD differed specifically in the SN where the ADHD group's neural profile suggests significantly less specificity in neural representations of spatial and verbal working memory. We highlight within-group classification as an innovative tool for understanding the neural mechanisms of how cognitive processes may deviate in clinical disorders, an important intermediary step towards improving translational psychiatry.

Dynamic and Static Amplitude of Low-Frequency Fluctuation Is a Potential Biomarker for Predicting Prognosis of Degenerative Cervical Myelopathy Patients: A Preliminary Resting-State fMRI Study

Objective

The aim of this study was to explore the clinical value of the static amplitude of low-frequency fluctuation (sALFF) and dynamic amplitude of low-frequency fluctuation (dALFF) in the identification of brain functional alterations in degenerative cervical myelopathy (DCM) patients.

Methods

Voxel-wise sALFF and dALFF of 47 DCM patients and 44 healthy controls were calculated using resting-state fMRI data, and an intergroup comparison was performed. The mean of sALFF or dALFF data were extracted within the resultant clusters and the correlation analysis of these data with the clinical measures was performed. Furthermore, whole-brain-wise and region-wise multivariate pattern analyses (MVPAs) were performed to classify DCM patients and healthy controls. sALFF and dALFF were used to predict the prognosis of DCM patients.

Results

The findings showed that (1) DCM patients exhibited higher sALFF within the left thalamus and putamen compared with that of the healthy controls. DCM patients also exhibited lower dALFF within bilateral postcentral gyrus compared with the healthy controls; (2) No significant correlations were observed between brain alterations and clinical measures through univariate correlation analysis; (3) sALFF (91%) and dALFF (95%) exhibited high accuracy in classifying the DCM patients and healthy controls; (4) Region-wise MVPA further revealed brain regions in which functional patterns were associated with prognosis in DCM patients. These regions were mainly located at the frontal lobe and temporal lobe.

Conclusion

In summary, sALFF and dALFF can be used to accurately reveal brain functional alterations in DCM patients. Furthermore, the multivariate approach is a more sensitive method in exploring neuropathology and establishing a prognostic biomarker for DCM compared with the conventional univariate method.

Identification and Therapeutic Outcome Prediction of Cervical Spondylotic Myelopathy Based on the Functional Connectivity From Resting-State Functional MRI Data: A Preliminary Machine Learning Study

Currently, strategies to diagnose patients and predict neurological recovery in cervical spondylotic myelopathy (CSM) using MR images of the cervical spine are urgently required. In light of this, this study aimed at exploring potential preoperative brain biomarkers that can be used to diagnose and predict neurological recovery in CSM patients using functional connectivity (FC) analysis of a resting-state functional MRI (rs-fMRI) data. Two independent datasets, including total of 53 patients with CSM and 47 age- and sex-matched healthy controls (HCs), underwent the preoperative rs-fMRI procedure. The FC was calculated from the automated anatomical labeling (AAL) template and used as features for machine learning analysis. After that, three analyses were used, namely, the classification of CSM patients from healthy adults using the support vector machine (SVM) within and across datasets, the prediction of preoperative neurological function in CSM patients via support vector regression (SVR) within and across datasets, and the prediction of neurological recovery in CSM patients via SVR within and across datasets. The results showed that CSM patients could be successfully identified from HCs with high classification accuracies (84.2% for dataset 1, 95.2% for dataset 2, and 73.0% for cross-site validation). Furthermore, the rs-FC combined with SVR could successfully predict the neurological recovery in CSM patients. Additionally, our results from cross-site validation analyses exhibited good reproducibility and generalization across the two datasets. Therefore, our findings provide preliminary evidence toward the development of novel strategies to predict neurological recovery in CSM patients using rs-fMRI and machine learning technique.

Multivariate pattern analysis links drug use severity to distributed cortical hypoactivity during emotional inhibitory control in opioid use disorder

Opioid use disorder (OUD) is characterized by emotional and cognitive impairements that are associated with poor treatment outcomes. The present study investigated the neural mechanism underlying emotion evaluation and inhibitory control using an affective go/no-go (AGN) task and its association with drug use severity and craving in patients with OUD. Twenty-six recently detoxified patients with OUD underwent functional magnetic resonance imaging (fMRI) while performing the AGN task that required response to frequently presented appetitive stimuli ("go") and inhibition of response to infrequently presented aversive stimuli ("no-go"). The fMRI session was immediately followed by an injection of extended-release opioid antagonist naltrexone (XR-NTX). Participants' opioid craving was assessed immediately before fMRI and 10 ± 2 days after XR-NTX injection. Multivariate pattern analysis (MVPA) showed that drug use severity was associated with distributed brain hypoactivity in response to aversive no-go stimuli, with particularly large negative contributions from the cognitive control and dorsal attention brain networks. While drug use severity and its associated MVPA brain response pattern were both correlated with opioid craving at baseline, only the brain response pattern predicted craving during XR-NTX treatment. Our findings point to widespread functional hypoactivity in the brain networks underlying emotional inhibitory control in OUD. Such a distributed pattern is consistent with the multifaceted nature of OUD, which affects multiple brain networks. It also highlights the utility of the multivariate approach in uncovering large-scale cortical substrates associated with clinical severity in complex psychiatric disorders and in predicting treatment response.

DarkASDNet: Classification of ASD on Functional MRI Using Deep Neural Network

Non-invasive whole-brain scans aid the diagnosis of neuropsychiatric disorder diseases such as autism, dementia, and brain cancer. The assessable analysis for autism spectrum disorders (ASD) is rationally challenging due to the limitations of publicly available datasets. For diagnostic or prognostic tools, functional Magnetic Resonance Imaging (fMRI) exposed affirmation to the biomarkers in neuroimaging research because of fMRI pickup inherent connectivity between the brain and regions. There are profound studies in ASD with introducing machine learning or deep learning methods that have manifested advanced steps for ASD predictions based on fMRI data. However, utmost antecedent models have an inadequacy in their capacity to manipulate performance metrics such as accuracy, precision, recall, and F1-score. To overcome these problems, we proposed an avant-garde DarkASDNet, which has the competence to extract features from a lower level to a higher level and bring out promising results. In this work, we considered 3D fMRI data to predict binary classification between ASD and typical control (TC). Firstly, we pre-processed the 3D fMRI data by adopting proper slice time correction and normalization. Then, we introduced a novel DarkASDNet which surpassed the benchmark accuracy for the classification of ASD. Our model's outcomes unveil that our proposed method established state-of-the-art accuracy of 94.70% to classify ASD vs. TC in ABIDE-I, NYU dataset. Finally, we contemplated our model by performing evaluation metrics including precision, recall, F1-score, ROC curve, and AUC score, and legitimize by distinguishing with recent literature descriptions to vindicate our outcomes. The proposed DarkASDNet architecture provides a novel benchmark approach for ASD classification using fMRI processed data.

Functional network estimation using multigraph learning with application to brain maturation study

Although most dramatic structural changes occur in the perinatal period, a growing body of evidences demonstrates that adolescence and early adulthood are also important for substantial neurodevelopment. We were thus motivated to explore brain development during puberty by evaluating functional connectivity network (FCN) differences between childhood and young adulthood using multi-paradigm task-based functional magnetic resonance imaging (fMRI) measurements. Different from conventional multigraph based FCN construction methods where the graph network was built independently for each modality/paradigm, we proposed a multigraph learning model in this work. It promises a better fitting to FCN construction by jointly estimating brain network from multi-paradigm fMRI time series, which may share common graph structures. To investigate the hub regions of the brain, we further conducted graph Fourier transform (GFT) to divide the fMRI BOLD time series of a node within the brain network into a range of frequencies. Then we identified the hub regions characterizing brain maturity through eigen-analysis of the low frequency components, which were believed to represent the organized structures shared by a large population. The proposed method was evaluated using both synthetic and real data, which demonstrated its effectiveness in extracting informative brain connectivity patterns. We detected 14 hub regions from the child group and 12 hub regions from the young adult group. We show the significance of these findings with a discussion of their functions and activation patterns as a function of age. In summary, our proposed method can extract brain connectivity network more accurately by considering the latent common structures between different fMRI paradigms, which are significant for both understanding brain development and recognizing population groups of different ages.

ISOMAP and machine learning algorithms for the construction of embedded functional connectivity networks of anatomically separated brain regions from resting state fMRI data of patients with Schizophrenia

We construct Functional Connectivity Networks (FCN) from resting state fMRI (rsfMRI) recordings towards the classification of brain activity between healthy and schizophrenic subjects using a publicly available dataset (the COBRE dataset) of 145 subjects (74 healthy controls and 71 schizophrenic subjects). First, we match the anatomy of the brain of each individual to the Desikan-Killiany brain atlas. Then, we use the conventional approach of correlating the parcellated time series to construct FCN and ISOMAP, a nonlinear manifold learning algorithm to produce low-dimensional embeddings of the correlation matrices. For the classification analysis, we computed five key local graph-theoretic measures of the FCN and used the LASSO and Random Forest (RF) algorithms for feature selection. For the classification we used standard linear Support Vector Machines. The classification performance is tested by a double cross-validation scheme (consisting of an outer and an inner loop of "Leave one out" cross-validation (LOOCV)). The standard cross-correlation methodology produced a classification rate of 73.1%, while ISOMAP resulted in 79.3%, thus providing a simpler model with a smaller number of features as chosen from LASSO and RF, namely the participation coefficient of the right thalamus and the strength of the right lingual gyrus.

Reliability of single-subject neural activation patterns in speech production tasks

Speech neuroimaging research targeting individual speakers could help elucidate differences that may be crucial to understanding speech disorders. However, this research necessitates reliable brain activation across multiple speech production sessions. In the present study, we evaluated the reliability of speech-related brain activity measured by functional magnetic resonance imaging data from twenty neuro-typical subjects who participated in two experiments involving reading aloud simple speech stimuli. Using traditional methods like the Dice and intraclass correlation coefficients, we found that most individuals displayed moderate to high reliability. We also found that a novel machine-learning subject classifier could identify these individuals by their speech activation patterns with 97% accuracy from among a dataset of seventy-five subjects. These results suggest that single-subject speech research would yield valid results and that investigations into the reliability of speech activation in people with speech disorders are warranted.

Multivariate Analysis of Structural and Functional Neuroimaging Can Inform Psychiatric Differential Diagnosis

Traditional psychiatric diagnosis has been overly reliant on either self-reported measures (introspection) or clinical rating scales (interviews). This produced the so-called explanatory gap with the bio-medical disciplines, such as neuroscience, which are supposed to deliver biological explanations of disease. In that context the neuro-biological and clinical assessment in psychiatry remained discrepant and incommensurable under conventional statistical frameworks. The emerging field of translational neuroimaging attempted to bridge the explanatory gap by means of simultaneous application of clinical assessment tools and functional magnetic resonance imaging, which also turned out to be problematic when analyzed with standard statistical methods. In order to overcome this problem our group designed a novel machine learning technique, multivariate linear method (MLM) which can capture convergent data from voxel-based morphometry, functional resting state and task-related neuroimaging and the relevant clinical measures. In this paper we report results from convergent cross-validation of biological signatures of disease in a sample of patients with schizophrenia as compared to depression. Our model provides evidence that the combination of the neuroimaging and clinical data in MLM analysis can inform the differential diagnosis in terms of incremental validity.

Machine Learning in Neuroimaging: A New Approach to Understand Acupuncture for Neuroplasticity

The effects of acupuncture facilitating neural plasticity for treating diseases have been identified by clinical and experimental studies. In the last two decades, the application of neuroimaging techniques in acupuncture research provided visualized evidence for acupuncture promoting neuroplasticity. Recently, the integration of machine learning (ML) and neuroimaging techniques becomes a focus in neuroscience and brings a new and promising approach to understand the facilitation of acupuncture on neuroplasticity at the individual level. This review is aimed at providing an overview of this rapidly growing field by introducing the commonly used ML algorithms in neuroimaging studies briefly and analyzing the characteristics of the acupuncture studies based on ML and neuroimaging, so as to provide references for future research.

Towards a brain-based predictome of mental illness

Neuroimaging-based approaches have been extensively applied to study mental illness in recent years and have deepened our understanding of both cognitively healthy and disordered brain structure and function. Recent advancements in machine learning techniques have shown promising outcomes for individualized prediction and characterization of patients with psychiatric disorders. Studies have utilized features from a variety of neuroimaging modalities, including structural, functional, and diffusion magnetic resonance imaging data, as well as jointly estimated features from multiple modalities, to assess patients with heterogeneous mental disorders, such as schizophrenia and autism. We use the term "predictome" to describe the use of multivariate brain network features from one or more neuroimaging modalities to predict mental illness. In the predictome, multiple brain network-based features (either from the same modality or multiple modalities) are incorporated into a predictive model to jointly estimate features that are unique to a disorder and predict subjects accordingly. To date, more than 650 studies have been published on subject-level prediction focusing on psychiatric disorders. We have surveyed about 250 studies including schizophrenia, major depression, bipolar disorder, autism spectrum disorder, attention-deficit hyperactivity disorder, obsessive-compulsive disorder, social anxiety disorder, posttraumatic stress disorder, and substance dependence. In this review, we present a comprehensive review of recent neuroimaging-based predictomic approaches, current trends, and common shortcomings and share our vision for future directions.

Resting-State Functional Network Scale Effects and Statistical Significance-Based Feature Selection in Machine Learning Classification

In recent years, functional brain network topological features have been widely used as classification features. Previous studies have found that network node scale differences caused by different network parcellation definitions significantly affect the structure of the constructed network and its topological properties. However, we still do not know how network scale differences affect the classification accuracy, performance of classification features, and effectiveness of the feature selection strategy using P values in terms of the machine learning method. This study used five scale parcellations, involving 90, 256, 497, 1003, and 1501 nodes. Three local properties of resting-state functional brain networks were selected (degree, betweenness centrality, and nodal efficiency), and the support vector machine method was used to construct classifiers to identify patients with major depressive disorder. We analyzed the impact of the five scales on classification accuracy. In addition, the effectiveness and redundancy of features obtained by the different scale parcellations were compared. Finally, traditional statistical significance (P value) was verified as a feature selection criterion. The results showed that the feature effectiveness of different scales was similar; in other words, parcellation with more regions did not provide more effective discriminative features. Nevertheless, parcellation with more regions did provide a greater quantity of discriminative features, which led to an improvement in the accuracy of the classification. However, due to the close distance between brain regions, the redundancy of parcellation with more regions was also greater. The traditional P value feature selection strategy is feasible with different scales, but our analysis showed that the traditional P < 0.05 threshold was too strict for feature selection. This study provides an important reference for the selection of network scales when applying topological properties of brain networks to machine learning methods.

Multivariate approaches improve the reliability and validity of functional connectivity and prediction of individual behaviors

Brain functional connectivity features can predict cognition and behavior at the level of the individual. Most studies measure univariate signals, correlating timecourses from the average of constituent voxels in each node. While straightforward, this approach overlooks the spatial patterns of voxel-wise signals within individual nodes. Given that multivariate spatial activity patterns across voxels can improve fMRI measures of mental representations, here we asked whether using voxel-wise timecourses can better characterize region-by-region interactions relative to univariate approaches. Using two fMRI datasets, the Human Connectome Project sample and a local test-retest sample, we measured multivariate functional connectivity with multivariate distance correlation and univariate connectivity with Pearson's correlation. We compared multivariate and univariate connectivity estimates, demonstrating that relative to univariate estimates, multivariate estimates exhibited higher reliability at both the edge-level and connectome-level, stronger prediction of individual differences, and greater sensitivity to brain states within individuals. Our findings suggest that multivariate estimates reliably provide more powerful information about an individual's functional brain organization and its relation to cognitive skills.

Evaluating the Prediction of Brain Maturity From Functional Connectivity After Motion Artifact Denoising

The ability to make individual-level predictions from neuroanatomy has the potential to be particularly useful in child development. Previously, resting-state functional connectivity (RSFC) MRI has been used to successfully predict maturity and diagnosis of typically and atypically developing individuals. Unfortunately, submillimeter head motion in the scanner produces systematic, distance-dependent differences in RSFC and may contaminate, and potentially facilitate, these predictions. Here, we evaluated individual age prediction with RSFC after stringent motion denoising. Using multivariate machine learning, we found that 57% of the variance in individual RSFC after motion artifact denoising was explained by age, while 4% was explained by residual effects of head motion. When RSFC data were not adequately denoised, 50% of the variance was explained by motion. Reducing motion-related artifact also revealed that prediction did not depend upon characteristics of functional connections previously hypothesized to mediate development (e.g., connection distance). Instead, successful age prediction relied upon sampling functional connections across multiple functional systems with strong, reliable RSFC within an individual. Our results demonstrate that RSFC across the brain is sufficiently robust to make individual-level predictions of maturity in typical development, and hence, may have clinical utility for the diagnosis and prognosis of individuals with atypical developmental trajectories.

Interaction of Developmental Venous Anomalies with Resting-State Functional MRI Measures

BACKGROUND AND PURPOSE

Functional MR imaging of the brain, used for both clinical and neuroscientific applications, relies on measuring fluctuations in blood oxygenation. Such measurements are susceptible to noise of vascular origin. The purpose of this study was to assess whether developmental venous anomalies, which are frequently observed normal variants, can bias fMRI measures by appearing as true neural signal.

MATERIALS AND METHODS

Large developmental venous anomalies (1 in each of 14 participants) were identified from a large neuroimaging cohort (n = 814). Resting-state fMRI data were decomposed using independent component analysis, a data-driven technique that creates distinct component maps representing aspects of either structured noise or true neural activity. We searched all independent components for maps that exhibited a spatial distribution of their signals following the topography of developmental venous anomalies.

RESULTS

Of the 14 developmental venous anomalies identified, 10 were clearly present in 17 fMRI independent components in total. While 9 (52.9%) of these 17 independent components were dominated by venous contributions and 2 (11.8%) by motion artifacts, 2 independent components (11.8%) showed partial neural signal contributions and 5 independent components (29.4%) unambiguously exhibited typical neural signal patterns.

CONCLUSIONS

Developmental venous anomalies can strongly resemble neural signal as measured by fMRI. They are thus a potential source of bias in fMRI analyses, especially when present in the cortex. This could impede interpretation of local activity in patients, such as in presurgical mapping. In scientific studies with large samples, developmental venous anomaly confounds could be mainly addressed using independent component analysis-based denoising.

Data Driven Classification Using fMRI Network Measures: Application to Schizophrenia

Using classification to identify biomarkers for various brain disorders has become a common practice among the functional MR imaging community. Typical classification pipeline includes taking the time series, extracting features from them, and using them to classify a set of patients and healthy controls. The most informative features are then presented as novel biomarkers. In this paper, we compared the results of single and double cross validation schemes on a cohort of 170 subjects with schizophrenia and healthy control subjects. We used graph theoretic measures as our features, comparing the use of functional and anatomical atlases to define nodes and the effect of prewhitening to remove autocorrelation trends. We found that double cross validation resulted in a 20% decrease in classification performance compared to single cross validation. The anatomical atlas resulted in higher classification results. Prewhitening resulted in a 10% boost in classification performance. Overall, a classification performance of 80% was obtained with a double-cross validation scheme using prewhitened time series and an anatomical brain atlas. However, reproducibility of classification within subjects across scans was surprisingly low and comparable to across subject classification rates, indicating that subject state during the short scan significantly influences the estimated features and classification performance.

Reproducibility of importance extraction methods in neural network based fMRI classification

Recent advances in machine learning allow faster training, improved performance and increased interpretability of classification techniques. Consequently, their application in neuroscience is rapidly increasing. While classification approaches have proved useful in functional magnetic resonance imaging (fMRI) studies, there are concerns regarding extraction, reproducibility and visualization of brain regions that contribute most significantly to the classification. We addressed these issues using an fMRI classification scheme based on neural networks and compared a set of methods for extraction of category-related voxel importances in three simulated and two empirical datasets. The simulation data revealed that the proposed scheme successfully detects spatially distributed and overlapping activation patterns upon successful classification. Application of the proposed classification scheme to two previously published empirical fMRI datasets revealed robust importance maps that extensively overlap with univariate maps but also provide complementary information. Our results demonstrate increased statistical power of importance maps compared to univariate approaches for both detection of overlapping patterns and patterns with weak univariate information.

Mind the prevalence rate: overestimating the clinical utility of psychiatric diagnostic classifiers

Currently, there is an intense pursuit of pathognomonic markers and diagnostic ('risk-based') classifiers of psychiatric conditions. Commonly, the epidemiological prevalence of the condition is not factored into the development of these classifiers. By not adjusting for prevalence, classifiers overestimate the potential of their clinical utility. As valid predictive values have critical implications in public health and allocation of resources, development of clinical classifiers should account for the prevalence of psychiatric conditions in both general and high-risk populations. We suggest that classifiers are most likely to be useful when targeting enriched populations.

Random support vector machine cluster analysis of resting-state fMRI in Alzheimer's disease

Early diagnosis is critical for individuals with Alzheimer's disease (AD) in clinical practice because its progress is irreversible. In the existing literature, support vector machine (SVM) has always been applied to distinguish between AD and healthy controls (HC) based on neuroimaging data. But previous studies have only used a single SVM to classify AD and HC, and the accuracy is not very high and generally less than 90%. The method of random support vector machine cluster was proposed to classify AD and HC in this paper. From the Alzheimer's Disease Neuroimaging Initiative database, the subjects including 25 AD individuals and 35 HC individuals were obtained. The classification accuracy could reach to 94.44% in the results. Furthermore, the method could also be used for feature selection and the accuracy could be maintained at the level of 94.44%. In addition, we could also find out abnormal brain regions (inferior frontal gyrus, superior frontal gyrus, precentral gyrus and cingulate cortex). It is worth noting that the proposed random support vector machine cluster could be a new insight to help the diagnosis of AD.

Classification of Autism Spectrum Disorder Using Random Support Vector Machine Cluster

Autism spectrum disorder (ASD) is mainly reflected in the communication and language barriers, difficulties in social communication, and it is a kind of neurological developmental disorder. Most researches have used the machine learning method to classify patients and normal controls, among which support vector machines (SVM) are widely employed. But the classification accuracy of SVM is usually low, due to the usage of a single SVM as classifier. Thus, we used multiple SVMs to classify ASD patients and typical controls (TC). Resting-state functional magnetic resonance imaging (fMRI) data of 46 TC and 61 ASD patients were obtained from the Autism Brain Imaging Data Exchange (ABIDE) database. Only 84 of 107 subjects are utilized in experiments because the translation or rotation of 7 TC and 16 ASD patients has surpassed ±2 mm or ±2°. Then the random SVM cluster was proposed to distinguish TC and ASD. The results show that this method has an excellent classification performance based on all the features. Furthermore, the accuracy based on the optimal feature set could reach to 96.15%. Abnormal brain regions could also be found, such as inferior frontal gyrus (IFG) (orbital and opercula part), hippocampus, and precuneus. It is indicated that the method of random SVM cluster may apply to the auxiliary diagnosis of ASD.

Sample heterogeneity in unipolar depression as assessed by functional connectivity analyses is dominated by general disease effects

OBJECTIVES

Combinations of resting-state fMRI and machine-learning techniques are increasingly employed to develop diagnostic models for mental disorders. However, little is known about the neurobiological heterogeneity of depression and diagnostic machine learning has mainly been tested in homogeneous samples. Our main objective was to explore the inherent structure of a diverse unipolar depression sample. The secondary objective was to assess, if such information can improve diagnostic classification.

MATERIALS AND METHODS

We analyzed data from 360 patients with unipolar depression and 360 non-depressed population controls, who were subdivided into two independent subsets. Cluster analyses (unsupervised learning) of functional connectivity were used to generate hypotheses about potential patient subgroups from the first subset. The relationship of clusters with demographical and clinical measures was assessed. Subsequently, diagnostic classifiers (supervised learning), which incorporated information about these putative depression subgroups, were trained.

RESULTS

Exploratory cluster analyses revealed two weakly separable subgroups of depressed patients. These subgroups differed in the average duration of depression and in the proportion of patients with concurrently severe depression and anxiety symptoms. The diagnostic classification models performed at chance level.

LIMITATIONS

It remains unresolved, if subgroups represent distinct biological subtypes, variability of continuous clinical variables or in part an overfitting of sparsely structured data.

CONCLUSIONS

Functional connectivity in unipolar depression is associated with general disease effects. Cluster analyses provide hypotheses about potential depression subtypes. Diagnostic models did not benefit from this additional information regarding heterogeneity.

A Computer-Aided Diagnosis System With EEG Based on the P3b Wave During an Auditory Odd-Ball Task in Schizophrenia

OBJECTIVE

To design a Computer-aided diagnosis (CAD) system using an optimized methodology over the P3b wave in order to objectively and accurately discriminate between healthy controls (HC) and schizophrenic subjects (SZ).

METHODS

We train, test, analyze, and compare various machine learning classification approaches optimized in terms of the correct classification rate (CCR), the degenerated Youden's index (DYI) and the area under the receiver operating curve (AUC). CAD system comprises five stages: electroencephalography (EEG) preprocessing, feature extraction, seven electrode groupings, discriminant feature selection, and binary classification.

RESULTS

With two optimal combinations of electrode grouping, filtering, feature selection algorithm, and classification machine, we get either a mean CCR = 93.42%, specificity = 0.9673, sensitivity = 0.8727, DYI = 0.9188, and AUC = 0.9567 (total-15 Hz-J5-MLP), or a mean CCR = 92.23%, specificity = 0.9499, sensitivity = 0.8838, DYI = 0.9162, and AUC = 0.9807 (right hemisphere-35 Hz-J5-SVM), which to our knowledge are higher than those available to date.

CONCLUSIONS

We have verified that a more restrictive low-pass filtering achieves higher CCR as compared to others at higher frequencies in the P3b wave. In addition, results validate previous hypothesis about the importance of the parietal-temporal region, associated with memory processing, allowing us to identify powerful {feature,electrode} pairs in the diagnosis of schizophrenia, achieving higher CCR and AUC in classification of both right and left Hemispheres, and parietal-temporal EEG signals, like, for instance, the {PSE, P4} pair (J5 and mutual information feature selection).

SIGNIFICANCE

Diagnosis of schizophrenia is made thoroughly by psychiatrists but as any human-based decision that has a subjective component. This CAD system provides the human expert with an objective complimentary measure to help him in diagnosing schizophrenia.

Combining Multiple Resting-State fMRI Features during Classification: Optimized Frameworks and Their Application to Nicotine Addiction

Machine learning techniques have been applied to resting-state fMRI data to predict neurological or neuropsychiatric disease states. Existing studies have used either a single type of resting-state feature or a few feature types (<4) in the prediction model. However, resting-state data can be processed in many different ways, yielding different feature types containing complementary and/or novel information, leaving uncertain the most informative features to provide to the classifier. In this study, multiple resting-state features were calculated from two main analytical categories: local measures and network measures. Feature selection was adopted using an optimized grid-search approach selecting top ranked features from statistical tests. We then tested three optimized frameworks: feature combination, kernel combination, and classifier combination, all using the support vector machine as an elementary classifier, to combine these resting-state feature types. When applied to nicotine addiction, with a cohort size of 100 smokers and 100 non-smokers, via a 10-fold cross-validation procedure, the feature combination and the classifier combination achieved an accuracy of 75.5%, while the kernel combination achieved a 73.0% accuracy; all three combination frameworks improved classification performance compared to the single feature type based results (best accuracy 70.5%). This study not only reveals the discriminative power of resting-state data, but also demonstrates the efficiency of combining multiple features from one data phenotype to improve classification performance.

How to Perform and Interpret Functional Magnetic Resonance Imaging Studies in Functional Gastrointestinal Disorders

Functional neuroimaging studies have revealed the importance of the role of cognitive and psychological factors and the dysregulation of the brain-gut axis in functional gastrointestinal disorder patients. Although only a small number of neuroimaging studies have been conducted in functional gastrointestinal disorder patients, and despite the fact that the neuroimaging technique requires a high level of knowledge, the technique still has a great deal of potential. The application of functional magnetic resonance imaging (fMRI) technique in functional gastrointestinal disorders should provide novel methods of diagnosing and treating patients. In this review, basic knowledge and technical/practical issues of fMRI will be introduced to clinicians.

Single subject prediction of brain disorders in neuroimaging: Promises and pitfalls

Neuroimaging-based single subject prediction of brain disorders has gained increasing attention in recent years. Using a variety of neuroimaging modalities such as structural, functional and diffusion MRI, along with machine learning techniques, hundreds of studies have been carried out for accurate classification of patients with heterogeneous mental and neurodegenerative disorders such as schizophrenia and Alzheimer's disease. More than 500 studies have been published during the past quarter century on single subject prediction focused on a multiple brain disorders. In the first part of this study, we provide a survey of more than 200 reports in this field with a focus on schizophrenia, mild cognitive impairment (MCI), Alzheimer's disease (AD), depressive disorders, autism spectrum disease (ASD) and attention-deficit hyperactivity disorder (ADHD). Detailed information about those studies such as sample size, type and number of extracted features and reported accuracy are summarized and discussed. To our knowledge, this is by far the most comprehensive review of neuroimaging-based single subject prediction of brain disorders. In the second part, we present our opinion on major pitfalls of those studies from a machine learning point of view. Common biases are discussed and suggestions are provided. Moreover, emerging trends such as decentralized data sharing, multimodal brain imaging, differential diagnosis, disease subtype classification and deep learning are also discussed. Based on this survey, there is extensive evidence showing the great potential of neuroimaging data for single subject prediction of various disorders. However, the main bottleneck of this exciting field is still the limited sample size, which could be potentially addressed by modern data sharing models such as the ones discussed in this paper. Emerging big data technologies and advanced data-intensive machine learning methodologies such as deep learning have coincided with an increasing need for accurate, robust and generalizable single subject prediction of brain disorders during an exciting time. In this report, we survey the past and offer some opinions regarding the road ahead.

Support Vector Machine Analysis of Functional Magnetic Resonance Imaging of Interoception Does Not Reliably Predict Individual Outcomes of Cognitive Behavioral Therapy in Panic Disorder with Agoraphobia

BACKGROUND

The approach to apply multivariate pattern analyses based on neuro imaging data for outcome prediction holds out the prospect to improve therapeutic decisions in mental disorders. Patients suffering from panic disorder with agoraphobia (PD/AG) often exhibit an increased perception of bodily sensations. The purpose of this investigation was to assess whether multivariate classification applied to a functional magnetic resonance imaging (fMRI) interoception paradigm can predict individual responses to cognitive behavioral therapy (CBT) in PD/AG.

METHODS

This analysis is based on pretreatment fMRI data during an interoceptive challenge from a multicenter trial of the German PANIC-NET. Patients with DSM-IV PD/AG were dichotomized as responders (n = 30) or non-responders (n = 29) based on the primary outcome (Hamilton Anxiety Scale Reduction ≥50%) after 6 weeks of CBT (2 h/week). fMRI parametric maps were used as features for response classification with linear support vector machines (SVM) with or without automated feature selection. Predictive accuracies were assessed using cross validation and permutation testing. The influence of methodological parameters and the predictive ability for specific interoception-related symptom reduction were further evaluated.

RESULTS

SVM did not reach sufficient overall predictive accuracies (38.0-54.2%) for anxiety reduction in the primary outcome. In the exploratory analyses, better accuracies (66.7%) were achieved for predicting interoception-specific symptom relief as an alternative outcome domain. Subtle information regarding this alternative response criterion but not the primary outcome was revealed by post hoc univariate comparisons.

CONCLUSION

In contrast to reports on other neurofunctional probes, SVM based on an interoception paradigm was not able to reliably predict individual response to CBT. Results speak against the clinical applicability of this technique.

Diagnostic classification of unipolar depression based on resting-state functional connectivity MRI: effects of generalization to a diverse sample

In small, selected samples, an approach combining resting-state functional connectivity MRI and multivariate pattern analysis has been able to successfully classify patients diagnosed with unipolar depression. Purposes of this investigation were to assess the generalizability of this approach to a large clinically more realistic sample and secondarily to assess the replicability of previously reported methodological feasibility in a more homogeneous subgroup with pronounced depressive symptoms. Two independent subsets were drawn from the depression and control cohorts of the BiDirect study, each with 180 patients with and 180 controls without depression. Functional connectivity either among regions covering the gray matter or selected regions with known alterations in depression was assessed by resting-state fMRI. Support vector machines with and without automated feature selection were used to train classifiers differentiating between individual patients and controls in the entire first subset as well as in the subgroup. Model parameters were explored systematically. The second independent subset was used for validation of successful models. Classification accuracies in the large, heterogeneous sample ranged from 45.0 to 56.1% (chance level 50.0%). In the subgroup with higher depression severity, three out of 90 models performed significantly above chance (60.8-61.7% at independent validation). In conclusion, common classification methods previously successful in small homogenous depression samples do not immediately translate to a more realistic population. Future research to develop diagnostic classification approaches in depression should focus on more specific clinical questions and consider heterogeneity, including symptom severity as an important factor.

Connectivity of the anterior insula differentiates participants with first-episode schizophrenia spectrum disorders from controls: a machine-learning study

BACKGROUND

Early diagnosis of schizophrenia could improve the outcomes and limit the negative effects of untreated illness. Although participants with schizophrenia show aberrant functional connectivity in brain networks, these between-group differences have a limited diagnostic utility. Novel methods of magnetic resonance imaging (MRI) analyses, such as machine learning (ML), may help bring neuroimaging from the bench to the bedside. Here, we used ML to differentiate participants with a first episode of schizophrenia-spectrum disorder (FES) from healthy controls based on resting-state functional connectivity (rsFC).

METHOD

We acquired resting-state functional MRI data from 63 patients with FES who were individually matched by age and sex to 63 healthy controls. We applied linear kernel support vector machines (SVM) to rsFC within the default mode network, the salience network and the central executive network.

RESULTS

The SVM applied to the rsFC within the salience network distinguished the FES from the control participants with an accuracy of 73.0% (p = 0.001), specificity of 71.4% and sensitivity of 74.6%. The classification accuracy was not significantly affected by medication dose, or by the presence of psychotic symptoms. The functional connectivity within the default mode or the central executive networks did not yield classification accuracies above chance level.

CONCLUSIONS

Seed-based functional connectivity maps can be utilized for diagnostic classification, even early in the course of schizophrenia. The classification was probably based on trait rather than state markers, as symptoms or medications were not significantly associated with classification accuracy. Our results support the role of the anterior insula/salience network in the pathophysiology of FES.

Multivariate pattern classification of pediatric Tourette syndrome using functional connectivity MRI

Tourette syndrome (TS) is a developmental neuropsychiatric disorder characterized by motor and vocal tics. Individuals with TS would benefit greatly from advances in prediction of symptom timecourse and treatment effectiveness. As a first step, we applied a multivariate method – support vector machine (SVM) classification – to test whether patterns in brain network activity, measured with resting state functional connectivity (RSFC) MRI, could predict diagnostic group membership for individuals. RSFC data from 42 children with TS (8–15 yrs) and 42 unaffected controls (age, IQ, in‐scanner movement matched) were included. While univariate tests identified no significant group differences, SVM classified group membership with ~70% accuracy (p < .001). We also report a novel adaptation of SVM binary classification that, in addition to an overall accuracy rate for the SVM, provides a confidence measure for the accurate classification of each individual. Our results support the contention that multivariate methods can better capture the complexity of some brain disorders, and hold promise for predicting prognosis and treatment outcome for individuals with TS.

Intrinsic brain activity as a diagnostic biomarker in children with benign epilepsy with centrotemporal spikes

Benign epilepsy with centrotemporal spikes (BECTS) is often associated with neural circuit dysfunction, particularly during the transient active state characterized by interictal epileptiform discharges (IEDs). Little is known, however, about the functional neural circuit abnormalities in BECTS without IEDs, or if such abnormalities could be used to differentiate BECTS patients without IEDs from healthy controls (HCs) for early diagnosis. To this end, we conducted resting-state functional magnetic resonance imaging (RS-fMRI) and simultaneous Electroencephalogram (EEG) in children with BECTS (n = 43) and age-matched HC (n = 28). The simultaneous EEG recordings distinguished BECTS with IEDs (n = 20) from without IEDs (n = 23). Intrinsic brain activity was measured in all three groups using the amplitude of low frequency fluctuation at rest. Compared to HC, BECTS patients with IEDs exhibited an intrinsic activity abnormality in the thalamus, suggesting that thalamic dysfunction could contribute to IED emergence while patients without IEDs exhibited intrinsic activity abnormalities in middle frontal gyrus and superior parietal gyrus. Using multivariate pattern classification analysis, we were able to differentiate BECTS without IEDs from HCs with 88.23% accuracy. BECTS without epileptic transients can be distinguished from HC and BECTS with IEDs by unique regional abnormalities in resting brain activity. Both transient abnormalities as reflected by IEDs and chronic abnormalities as reflected by RS-fMRI may contribute to BECTS development and expression. Intrinsic brain activity and multivariate pattern classification techniques are promising tools to diagnose and differentiate BECTS syndromes. Hum Brain Mapp 36:3878-3889, 2015. © 2015 Wiley Periodicals, Inc.

Diagnostic classification of schizophrenia patients on the basis of regional reward-related FMRI signal patterns

Functional neuroimaging has provided evidence for altered function of mesolimbic circuits implicated in reward processing, first and foremost the ventral striatum, in patients with schizophrenia. While such findings based on significant group differences in brain activations can provide important insights into the pathomechanisms of mental disorders, the use of neuroimaging results from standard univariate statistical analysis for individual diagnosis has proven difficult. In this proof of concept study, we tested whether the predictive accuracy for the diagnostic classification of schizophrenia patients vs. healthy controls could be improved using multivariate pattern analysis (MVPA) of regional functional magnetic resonance imaging (fMRI) activation patterns for the anticipation of monetary reward. With a searchlight MVPA approach using support vector machine classification, we found that the diagnostic category could be predicted from local activation patterns in frontal, temporal, occipital and midbrain regions, with a maximal cluster peak classification accuracy of 93% for the right pallidum. Region-of-interest based MVPA for the ventral striatum achieved a maximal cluster peak accuracy of 88%, whereas the classification accuracy on the basis of standard univariate analysis reached only 75%. Moreover, using support vector regression we could additionally predict the severity of negative symptoms from ventral striatal activation patterns. These results show that MVPA can be used to substantially increase the accuracy of diagnostic classification on the basis of task-related fMRI signal patterns in a regionally specific way.

Toward literature-based feature selection for diagnostic classification: a meta-analysis of resting-state fMRI in depression

Information derived from functional magnetic resonance imaging (fMRI) during wakeful rest has been introduced as a candidate diagnostic biomarker in unipolar major depressive disorder (MDD). Multiple reports of resting state fMRI in MDD describe group effects. Such prior knowledge can be adopted to pre-select potentially discriminating features for diagnostic classification models with the aim to improve diagnostic accuracy. Purpose of this analysis was to consolidate spatial information about alterations of spontaneous brain activity in MDD, primarily to serve as feature selection for multivariate pattern analysis techniques (MVPA). Thirty two studies were included in final analyses. Coordinates extracted from the original reports were assigned to two categories based on directionality of findings. Meta-analyses were calculated using the non-additive activation likelihood estimation approach with coordinates organized by subject group to account for non-independent samples. Converging evidence revealed a distributed pattern of brain regions with increased or decreased spontaneous activity in MDD. The most distinct finding was hyperactivity/hyperconnectivity presumably reflecting the interaction of cortical midline structures (posterior default mode network components including the precuneus and neighboring posterior cingulate cortices associated with self-referential processing and the subgenual anterior cingulate and neighboring medial frontal cortices) with lateral prefrontal areas related to externally-directed cognition. Other areas of hyperactivity/hyperconnectivity include the left lateral parietal cortex, right hippocampus and right cerebellum whereas hypoactivity/hypoconnectivity was observed mainly in the left temporal cortex, the insula, precuneus, superior frontal gyrus, lentiform nucleus and thalamus. Results are made available in two different data formats to be used as spatial hypotheses in future studies, particularly for diagnostic classification by MVPA.

Contributions and complexities from the use of in vivo animal models to improve understanding of human neuroimaging signals

Many of the major advances in our understanding of how functional brain imaging signals relate to neuronal activity over the previous two decades have arisen from physiological research studies involving experimental animal models. This approach has been successful partly because it provides opportunities to measure both the hemodynamic changes that underpin many human functional brain imaging techniques and the neuronal activity about which we wish to make inferences. Although research into the coupling of neuronal and hemodynamic responses using animal models has provided a general validation of the correspondence of neuroimaging signals to specific types of neuronal activity, it is also highlighting the key complexities and uncertainties in estimating neural signals from hemodynamic markers. This review will detail how research in animal models is contributing to our rapidly evolving understanding of what human neuroimaging techniques tell us about neuronal activity. It will highlight emerging issues in the interpretation of neuroimaging data that arise from in vivo research studies, for example spatial and temporal constraints to neuroimaging signal interpretation, or the effects of disease and modulatory neurotransmitters upon neurovascular coupling. We will also give critical consideration to the limitations and possible complexities of translating data acquired in the typical animals models used in this area to the arena of human fMRI. These include the commonplace use of anesthesia in animal research studies and the fact that many neuropsychological questions that are being actively explored in humans have limited homologs within current animal models for neuroimaging research. Finally we will highlighting approaches, both in experimental animals models (e.g. imaging in conscious, behaving animals) and human studies (e.g. combined fMRI-EEG), that mitigate against these challenges.

Establishing the bidirectional relationship between depression and subclinical arteriosclerosis--rationale, design, and characteristics of the BiDirect Study

BACKGROUND

Depression and cardiovascular diseases due to arteriosclerosis are both frequent and impairing conditions. Depression and (subclinical) arteriosclerosis appear to be related in a bidirectional way, and it is plausible to assume a partly joint causal relationship. However, the biological mechanisms and the behavioral pathways that lead from depression to arteriosclerosis and vice versa remain to be exactly determined.

METHODS/DESIGN

This study protocol describes the rationale and design of the prospective BiDirect Study that aims at investigating the mutual relationship between depression and (subclinical) arteriosclerosis. BiDirect is scheduled to follow-up three distinct cohorts of individuals ((i) patients with acute depression (N = 999), (ii) patients after an acute cardiac event (N = 347), and (iii) reference subjects from the general population (N = 912)). Over the course of 12 years, four personal examinations are planned to be conducted. The core examination program, which will remain identical across follow-ups, comprises a personal interview (e.g. medical diagnoses, health care utilization, lifestyle and risk behavior), a battery of self-administered questionnaires (e.g. depressive symptoms, readiness to change health behavior, perceived health-related quality of life), sensory (e.g. olfaction, pain) and neuropsychological (e.g. memory, executive functions, emotional processing, manual dexterity) assessments, anthropometry, body impedance measurement, a clinical work-up regarding the vascular status (e.g. electrocardiogram, blood pressure, intima media thickness), the taking of blood samples (serum and plasma, DNA), and structural and functional resonance imaging of the brain (e.g. diffusion tensor imaging, resting-state, emotional faces processing). The present report includes BiDirect-Baseline, the first data collection wave.

DISCUSSION

Due to its prospective character, the integration of three distinct cohorts, the long follow-up time window, the diligent diagnosis of depression taking depression subtypes into account, the consideration of relevant comorbidities and risk factors, the assessment of indicators of (subclinical) arteriosclerosis in different vascular territories, and the structural and functional brain imaging that is performed for a large number of participants, the BiDirect Study represents an innovative approach that combines population-based cohorts with sophisticated clinical work-up methods and that holds the potential to overcome many of the drawbacks characterizing earlier investigations.

Disease prediction based on functional connectomes using a scalable and spatially-informed support vector machine

Substantial evidence indicates that major psychiatric disorders are associated with distributed neural dysconnectivity, leading to a strong interest in using neuroimaging methods to accurately predict disorder status. In this work, we are specifically interested in a multivariate approach that uses features derived from whole-brain resting state functional connectomes. However, functional connectomes reside in a high dimensional space, which complicates model interpretation and introduces numerous statistical and computational challenges. Traditional feature selection techniques are used to reduce data dimensionality, but are blind to the spatial structure of the connectomes. We propose a regularization framework where the 6-D structure of the functional connectome (defined by pairs of points in 3-D space) is explicitly taken into account via the fused Lasso or the GraphNet regularizer. Our method only restricts the loss function to be convex and margin-based, allowing non-differentiable loss functions such as the hinge-loss to be used. Using the fused Lasso or GraphNet regularizer with the hinge-loss leads to a structured sparse support vector machine (SVM) with embedded feature selection. We introduce a novel efficient optimization algorithm based on the augmented Lagrangian and the classical alternating direction method, which can solve both fused Lasso and GraphNet regularized SVM with very little modification. We also demonstrate that the inner subproblems of the algorithm can be solved efficiently in analytic form by coupling the variable splitting strategy with a data augmentation scheme. Experiments on simulated data and resting state scans from a large schizophrenia dataset show that our proposed approach can identify predictive regions that are spatially contiguous in the 6-D "connectome space," offering an additional layer of interpretability that could provide new insights about various disease processes.