Neuroimaging-based classification of PTSD using data-driven computational approaches: A multisite big data study from the ENIGMA-PGC PTSD consortium.

BACKGROUND: Recent advances in data-driven computational approaches have been helpful in devising tools to objectively diagnose psychiatric disorders. However, current machine learning studies limited to small homogeneous samples, different methodologies, and different imaging collection protocols, limit the ability to directly compare and generalize their results. Here we aimed to classify individuals with PTSD versus controls and assess the generalizability using a large heterogeneous brain datasets from the ENIGMA-PGC PTSD Working group. METHODS: We analyzed brain MRI data from 3,477 structural-MRI; 2,495 resting state-fMRI; and 1,952 diffusion-MRI. First, we identified the brain features that best distinguish individuals with PTSD from controls using traditional machine learning methods. Second, we assessed the utility of the denoising variational autoencoder (DVAE) and evaluated its classification performance. Third, we assessed the generalizability and reproducibility of both models using leave-one-site-out cross-validation procedure for each modality. RESULTS: We found lower performance in classifying PTSD vs. controls with data from over 20 sites (60 % test AUC for s-MRI, 59 % for rs-fMRI and 56 % for d-MRI), as compared to other studies run on single-site data. The performance increased when classifying PTSD from HC without trauma history in each modality (75 % AUC). The classification performance remained intact when applying the DVAE framework, which reduced the number of features. Finally, we found that the DVAE framework achieved better generalization to unseen datasets compared with the traditional machine learning frameworks, albeit performance was slightly above chance. CONCLUSION: These results have the potential to provide a baseline classification performance for PTSD when using large scale neuroimaging datasets. Our findings show that the control group used can heavily affect classification performance. The DVAE framework provided better generalizability for the multi-site data. This may be more significant in clinical practice since the neuroimaging-based diagnostic DVAE classification models are much less site-specific, rendering them more generalizable.

Anomalous network architecture of the resting brain in children who stutter.

PURPOSE: We combined a large longitudinal neuroimaging dataset that includes children who do and do not stutter and a whole-brain network analysis in order to examine the intra- and inter-network connectivity changes associated with stuttering. Additionally, we asked whether whole brain connectivity patterns observed at the initial year of scanning could predict persistent stuttering in later years. METHODS: A total of 224 high-quality resting state fMRI scans collected from 84 children (42 stuttering, 42 controls) were entered into an independent component analysis (ICA), yielding a number of distinct network connectivity maps ("components") as well as expression scores for each component that quantified the degree to which it is expressed for each child. These expression scores were compared between stuttering and control groups' first scans. In a second analysis, we examined whether the components that were most predictive of stuttering status also predicted persistence in stuttering. RESULTS: Stuttering status, as well as stuttering persistence, were associated with aberrant network connectivity involving the default mode network and its connectivity with attention, somatomotor, and frontoparietal networks. The results suggest developmental alterations in the balance of integration and segregation of large-scale neural networks that support proficient task performance including fluent speech motor control. CONCLUSIONS: This study supports the view that stuttering is a complex neurodevelopmental disorder and provides comprehensive brain network maps that substantiate past theories emphasizing the importance of considering situational, emotional, attentional and linguistic factors in explaining the basis for stuttering onset, persistence, and recovery.

ALTERED DEFAULT MODE NETWORK (DMN) RESTING STATE FUNCTIONAL CONNECTIVITY FOLLOWING A MINDFULNESS-BASED EXPOSURE THERAPY FOR POSTTRAUMATIC STRESS DISORDER (PTSD) IN COMBAT VETERANS OF AFGHANISTAN AND IRAQ.

BACKGROUND: Recent studies suggest that mindfulness may be an effective component for posttraumatic stress disorder (PTSD) treatment. Mindfulness involves practice in volitional shifting of attention from "mind wandering" to present-moment attention to sensations, and cultivating acceptance. We examined potential neural correlates of mindfulness training using a novel group therapy (mindfulness-based exposure therapy (MBET)) in combat veterans with PTSD deployed to Afghanistan (OEF) and/or Iraq (OIF). METHODS: Twenty-three male OEF/OIF combat veterans with PTSD were treated with a mindfulness-based intervention (N = 14) or an active control group therapy (present-centered group therapy (PCGT), N = 9). Pre-post therapy functional magnetic resonance imaging (fMRI, 3 T) examined resting-state functional connectivity (rsFC) in default mode network (DMN) using posterior cingulate cortex (PCC) and ventral medial prefrontal cortex (vmPFC) seeds, and salience network (SN) with anatomical amygdala seeds. PTSD symptoms were assessed at pre- and posttherapy with Clinician Administered PTSD Scale (CAPS). RESULTS: Patients treated with MBET had reduced PTSD symptoms (effect size d = 0.92) but effect was not significantly different from PCGT (d = 0.46). Increased DMN rsFC (PCC seed) with dorsolateral dorsolateral prefrontal cortex (DLPFC) regions and dorsal anterior cingulate cortex (ACC) regions associated with executive control was seen following MBET. A group × time interaction found MBET showed increased connectivity with DLPFC and dorsal ACC following therapy; PCC-DLPFC connectivity was correlated with improvement in PTSD avoidant and hyperarousal symptoms. CONCLUSIONS: Increased connectivity between DMN and executive control regions following mindfulness training could underlie increased capacity for volitional shifting of attention. The increased PCC-DLPFC rsFC following MBET was related to PTSD symptom improvement, pointing to a potential therapeutic mechanism of mindfulness-based therapies.

Growth Charting of Brain Connectivity Networks and the Identification of Attention Impairment in Youth.

IMPORTANCE: Intrinsic connectivity networks (ICNs), important units of brain functional organization, demonstrate substantial maturation during youth. In addition, interrelationships between ICNs have been reliably implicated in attention performance. It is unknown whether alterations in ICN maturational profiles can reliably detect impaired attention functioning in youth. OBJECTIVE: To use a network growth charting approach to investigate the association between alterations in ICN maturation and attention performance. DESIGN, SETTING, AND PARTICIPANTS: Data were obtained from the publicly available Philadelphia Neurodevelopmental Cohort, a prospective, population-based sample of 9498 youths who underwent genomic testing, neurocognitive assessment, and neuroimaging. Data collection was conducted at an academic and children's hospital health care network between November 1, 2009, and November 30, 2011, and data analysis was conducted between February 1, 2015, and January 15, 2016. MAIN OUTCOMES AND MEASURES: Statistical associations between deviations from normative network growth were assessed as well as 2 main outcome measures: accuracy during the Penn Continuous Performance Test and diagnosis with attention-deficit/hyperactivity disorder. RESULTS: Of the 9498 individuals identified, 1000 youths aged 8 to 22 years underwent brain imaging. A sample of 519 youths who met quality control criteria entered analysis, of whom 25 (4.8%) met criteria for attention-deficit/hyperactivity disorder. The mean (SD) age of the youth was 15.7 (3.1) years, and 223 (43.0%) were male. Participants' patterns of deviations from normative maturational trajectories were indicative of sustained attention functioning (R2 = 24%; F6,512 = 26.89; P < 2.2 × 10-16). Moreover, these patterns were found to be a reliable biomarker of severe attention impairment (peak receiver operating characteristic curve measured by area under the curve, 79.3%). In particular, a down-shifted pattern of ICN maturation (shallow maturation), rather than a right-shifted pattern (lagged maturation), was implicated in reduced attention performance (Akaike information criterion relative likelihood, 3.22 × 1026). Finally, parallel associations between ICN dysmaturation and diagnosis of attention-deficit/hyperactivity disorder were identified. CONCLUSIONS AND RELEVANCE: Growth charting methods are widely used to assess the development of physical or other biometric characteristics, such as weight and head circumference. To date, this is the first demonstration that this method can be extended to development of functional brain networks to identify clinically relevant conditions, such as dysfunction of sustained attention.

Childhood Cumulative Risk Exposure and Adult Amygdala Volume and Function.

Considerable work indicates that early cumulative risk exposure is aversive to human development, but very little research has examined the neurological underpinnings of these robust findings. This study investigates amygdala volume and reactivity to facial stimuli among adults (mean 23.7 years of age, n = 54) as a function of cumulative risk exposure during childhood (9 and 13 years of age). In addition, we test to determine whether expected cumulative risk elevations in amygdala volume would mediate functional reactivity of the amygdala during socioemotional processing. Risks included substandard housing quality, noise, crowding, family turmoil, child separation from family, and violence. Total and left hemisphere adult amygdala volumes were positively related to cumulative risk exposure during childhood. The links between childhood cumulative risk exposure and elevated amygdala responses to emotionally neutral facial stimuli in adulthood were mediated by the corresponding amygdala volumes. Cumulative risk exposure in later adolescence (17 years of age), however, was unrelated to subsequent adult amygdala volume or function. Physical and socioemotional risk exposures early in life appear to alter amygdala development, rendering adults more reactive to ambiguous stimuli such as neutral faces. These stress-related differences in childhood amygdala development might contribute to the well-documented psychological distress as a function of early risk exposure.

Childhood Poverty Predicts Adult Amygdala and Frontal Activity and Connectivity in Response to Emotional Faces.

Childhood poverty negatively impacts physical and mental health in adulthood. Altered brain development in response to social and environmental factors associated with poverty likely contributes to this effect, engendering maladaptive patterns of social attribution and/or elevated physiological stress. In this fMRI study, we examined the association between childhood poverty and neural processing of social signals (i.e., emotional faces) in adulthood. Fifty-two subjects from a longitudinal prospective study recruited as children, participated in a brain imaging study at 23-25 years of age using the Emotional Faces Assessment Task. Childhood poverty, independent of concurrent adult income, was associated with higher amygdala and medial prefrontal cortical (mPFC) responses to threat vs. happy faces. Also, childhood poverty was associated with decreased functional connectivity between left amygdala and mPFC. This study is unique, because it prospectively links childhood poverty to emotional processing during adulthood, suggesting a candidate neural mechanism for negative social-emotional bias. Adults who grew up poor appear to be more sensitive to social threat cues and less sensitive to positive social cues.

Modality-spanning deficits in attention-deficit/hyperactivity disorder in functional networks, gray matter, and white matter.

Previous neuroimaging investigations in attention-deficit/hyperactivity disorder (ADHD) have separately identified distributed structural and functional deficits, but interconnections between these deficits have not been explored. To unite these modalities in a common model, we used joint independent component analysis, a multivariate, multimodal method that identifies cohesive components that span modalities. Based on recent network models of ADHD, we hypothesized that altered relationships between large-scale networks, in particular, default mode network (DMN) and task-positive networks (TPNs), would co-occur with structural abnormalities in cognitive regulation regions. For 756 human participants in the ADHD-200 sample, we produced gray and white matter volume maps with voxel-based morphometry, as well as whole-brain functional connectomes. Joint independent component analysis was performed, and the resulting transmodal components were tested for differential expression in ADHD versus healthy controls. Four components showed greater expression in ADHD. Consistent with our a priori hypothesis, we observed reduced DMN-TPN segregation co-occurring with structural abnormalities in dorsolateral prefrontal cortex and anterior cingulate cortex, two important cognitive control regions. We also observed altered intranetwork connectivity in DMN, dorsal attention network, and visual network, with co-occurring distributed structural deficits. There was strong evidence of spatial correspondence across modalities: For all four components, the impact of the respective component on gray matter at a region strongly predicted the impact on functional connectivity at that region. Overall, our results demonstrate that ADHD involves multiple, cohesive modality spanning deficits, each one of which exhibits strong spatial overlap in the pattern of structural and functional alterations.

Effects of oxycodone on brain responses to emotional images.

RATIONALE: Evidence from animal and human studies suggests that opiate drugs decrease emotional responses to negative stimuli and increase responses to positive stimuli. Such emotional effects may motivate misuse of oxycodone (OXY), a widely abused opiate. Yet, we know little about how OXY affects neural circuits underlying emotional processing in humans. OBJECTIVE: We examined effects of OXY on brain activity during presentation of positive and negative visual emotional stimuli. We predicted that OXY would decrease amygdala activity to negative stimuli and increase ventral striatum (VS) activity to positive stimuli. Secondarily, we examined the effects of OXY on other emotional network regions on an exploratory basis. METHODS: In a three-session study, healthy adults (N = 17) received placebo, 10 and 20 mg OXY under counterbalanced, double-blind conditions. At each session, participants completed subjective and cardiovascular measures and underwent functional MRI (fMRI) scanning while completing two emotional response tasks. RESULTS: Our emotional tasks reliably activated emotional network areas. OXY produced subjective effects but did not alter either behavioral responses to emotional stimuli or activity in our primary areas of interest. OXY did decrease right medial orbitofrontal cortex (MOFC) responses to happy faces. CONCLUSIONS: Contrary to our expectations, OXY did not affect behavioral or neural responses to emotional stimuli in our primary areas of interest. Further, the effects of OXY in the MOFC would be more consistent with a decrease in value for happy faces. This may indicate that healthy adults do not receive emotional benefits from opiates, or the pharmacological actions of OXY differ from other opiates.

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.

Disrupted network architecture of the resting brain in attention-deficit/hyperactivity disorder.

BACKGROUND: Attention-deficit/hyperactivity disorder (ADHD) is one of the most prevalent psychiatric disorders of childhood. Neuroimaging investigations of ADHD have traditionally sought to detect localized abnormalities in discrete brain regions. Recent years, however, have seen the emergence of complementary lines of investigation into distributed connectivity disturbances in ADHD. Current models emphasize abnormal relationships between default network-involved in internally directed mentation and lapses of attention-and task positive networks, especially ventral attention network. However, studies that comprehensively investigate interrelationships between large-scale networks in ADHD remain relatively rare. METHODS: Resting state functional magnetic resonance imaging scans were obtained from 757 participants at seven sites in the ADHD-200 multisite sample. Functional connectomes were generated for each subject, and interrelationships between seven large-scale brain networks were examined with network contingency analysis. RESULTS: ADHD brains exhibited altered resting state connectivity between default network and ventral attention network [P < 0.0001, false discovery rate (FDR)-corrected], including prominent increased connectivity (more specifically, diminished anticorrelation) between posterior cingulate cortex in default network and right anterior insula and supplementary motor area in ventral attention network. There was distributed hypoconnectivity within default network (P = 0.009, FDR-corrected), and this network also exhibited significant alterations in its interconnections with several other large-scale networks. Additionally, there was pronounced right lateralization of aberrant default network connections. CONCLUSIONS: Consistent with existing theoretical models, these results provide evidence that default network-ventral attention network interconnections are a key locus of dysfunction in ADHD. Moreover, these findings contribute to growing evidence that distributed dysconnectivity within and between large-scale networks is present in ADHD.

Volitional regulation of emotions produces distributed alterations in connectivity between visual, attention control, and default networks.

The ability to volitionally regulate emotions is critical to health and well-being. While patterns of neural activation during emotion regulation have been well characterized, patterns of connectivity between regions remain less explored. It is increasingly recognized that the human brain is organized into large-scale intrinsic connectivity networks (ICNs) whose interrelationships are altered in characteristic ways during psychological tasks. In this fMRI study of 54 healthy individuals, we investigated alterations in connectivity within and between ICNs produced by the emotion regulation strategy of reappraisal. In order to gain a comprehensive picture of connectivity changes, we utilized connectomic psychophysiological interactions (PPI), a whole-brain generalization of standard single-seed PPI methods. In particular, we quantified PPI connectivity pair-wise across 837 ROIs placed throughout the cortex. We found that compared to maintaining one's emotional responses, engaging in reappraisal produced robust and distributed alterations in functional connections involving visual, dorsal attention, frontoparietal, and default networks. Visual network in particular increased connectivity with multiple ICNs including dorsal attention and default networks. We interpret these findings in terms of the role of these networks in mediating critical constituent processes in emotion regulation, including visual processing, stimulus salience, attention control, and interpretation and contextualization of stimuli. Our results add a new network perspective to our understanding of the neural underpinnings of emotion regulation, and highlight that connectomic methods can play a valuable role in comprehensively investigating modulation of connectivity across task conditions.

SCALABLE FUSED LASSO SVM FOR CONNECTOME-BASED DISEASE PREDICTION.

There is substantial interest in developing machine-based methods that reliably distinguish patients from healthy controls using high dimensional correlation maps known as functional connectomes (FC's) generated from resting state fMRI. To address the dimensionality of FC's, the current body of work relies on feature selection techniques that are blind to the spatial structure of the data. In this paper, we propose to use the fused Lasso regularized support vector machine to explicitly account for the 6-D structure of the FC (defined by pairs of points in 3-D brain space). In order to solve the resulting nonsmooth and large-scale optimization problem, we introduce a novel and scalable algorithm based on the alternating direction method. Experiments on real resting state scans show that our approach can recover results that are more neuroscientifically informative than previous methods.

Effects of childhood poverty and chronic stress on emotion regulatory brain function in adulthood.

Childhood poverty has pervasive negative physical and psychological health sequelae in adulthood. Exposure to chronic stressors may be one underlying mechanism for childhood poverty-health relations by influencing emotion regulatory systems. Animal work and human cross-sectional studies both suggest that chronic stressor exposure is associated with amygdala and prefrontal cortex regions important for emotion regulation. In this longitudinal functional magnetic resonance imaging study of 49 participants, we examined associations between childhood poverty at age 9 and adult neural circuitry activation during emotion regulation at age 24. To test developmental timing, concurrent, adult income was included as a covariate. Adults with lower family income at age 9 exhibited reduced ventrolateral and dorsolateral prefrontal cortex activity and failure to suppress amygdala activation during effortful regulation of negative emotion at age 24. In contrast to childhood income, concurrent adult income was not associated with neural activity during emotion regulation. Furthermore, chronic stressor exposure across childhood (at age 9, 13, and 17) mediated the relations between family income at age 9 and ventrolateral and dorsolateral prefrontal cortex activity at age 24. The findings demonstrate the significance of childhood chronic stress exposures in predicting neural outcomes during emotion regulation in adults who grew up in poverty.

The caudate signals bad reputation during trust decisions.

The ability to initiate and sustain trust is critical to health and well-being. Willingness to trust is in part determined by the reputation of the putative trustee, gained via direct interactions or indirectly through word of mouth. Few studies have examined how the reputation of others is instantiated in the brain during trust decisions. Here we use an event-related functional MRI (fMRI) design to examine what neural signals correspond to experimentally manipulated reputations acquired in direct interactions during trust decisions. We hypothesized that the caudate (dorsal striatum) and putamen (ventral striatum) and amygdala would signal differential reputations during decision-making. Twenty-nine healthy adults underwent fMRI scanning while completing an iterated Trust Game as trusters with three fictive trustee partners who had different tendencies to reciprocate (i.e., likelihood of rewarding the truster), which were learned over multiple exchanges with real-time feedback. We show that the caudate (both left and right) signals reputation during trust decisions, such that caudate is more active to partners with two types of "bad" reputations, either indifferent partners (who reciprocate 50% of the time) or unfair partners (who reciprocate 25% of the time), than to those with "good" reputations (who reciprocate 75% of the time). Further, individual differences in caudate activity related to biases in trusting behavior in the most uncertain situation, i.e. when facing an indifferent partner. We also report on other areas that were activated by reputation at p < 0.05 whole brain corrected. Our findings suggest that the caudate is involved in signaling and integrating reputations gained through experience into trust decisions, demonstrating a neural basis for this key social process.

Distributed effects of methylphenidate on the network structure of the resting brain: a connectomic pattern classification analysis.

Methylphenidate is a psychostimulant medication that produces improvements in functions associated with multiple neurocognitive systems. To investigate the potentially distributed effects of methylphenidate on the brain's intrinsic network architecture, we coupled resting state imaging with multivariate pattern classification. In a within-subject, double-blind, placebo-controlled, randomized, counterbalanced, cross-over design, 32 healthy human volunteers received either methylphenidate or placebo prior to two fMRI resting state scans separated by approximately one week. Resting state connectomes were generated by placing regions of interest at regular intervals throughout the brain, and these connectomes were submitted for support vector machine analysis. We found that methylphenidate produces a distributed, reliably detected, multivariate neural signature. Methylphenidate effects were evident across multiple resting state networks, especially visual, somatomotor, and default networks. Methylphenidate reduced coupling within visual and somatomotor networks. In addition, default network exhibited decoupling with several task positive networks, consistent with methylphenidate modulation of the competitive relationship between these networks. These results suggest that connectivity changes within and between large-scale networks are potentially involved in the mechanisms by which methylphenidate improves attention functioning.

Aberrant reward center response to partner reputation during a social exchange game in generalized social phobia.

OBJECTIVE: Generalized social anxiety disorder (GSAD) is characterized by excessive fear of public scrutiny and reticence in social engagement. Previous studies have probed the neural basis of GSAD often using static, noninteractive stimuli (e.g., face photographs) and have identified dysfunction in fear circuitry. We sought to investigate brain-based dysfunction in GSAD during more real-world, dynamic social interactions, focusing on the role of reward-related regions that are implicated in social decision-making. METHODS: Thirty-six healthy individuals (healthy control [HC]) and 36 individuals with GSAD underwent functional magnetic resonance imaging (fMRI) scanning while participating in a behavioral economic game ("Trust Game") involving iterative exchanges with fictive partners who acquire differential reputations for reciprocity. We investigated brain responses to reciprocation of trust in one's social partner, and how these brain responses are modulated by partner reputation for repayment. RESULTS: In both HC and GSAD, receipt of reciprocity robustly engaged ventral striatum, a region implicated in reward. In HC, striatal responses to reciprocity were specific to partners who have consistently returned the investment ("cooperative partners"), and were absent for partners who lack a cooperative reputation. In GSAD, modulation of striatal responses by partner reputation was absent. Social anxiety severity predicted diminished responses to cooperative partners. CONCLUSION: These results suggest abnormalities in GSAD in reward-related striatal mechanisms that may be important for the initiation, valuation, and maintenance of cooperative social relationships. Moreover, this study demonstrates that dynamic, interactive task paradigms derived from economics can help illuminate novel mechanisms of pathology in psychiatric illnesses in which social dysfunction is a cardinal feature.

Neural substrates of alcohol-induced smoking urge in heavy drinking nondaily smokers.

A strong link exists between cigarette smoking and alcohol use, which may be explained by the experimental observation that alcohol ingestion promotes cigarette craving and precipitates smoking. At the neuroanatomic level, it is unclear where and how alcohol exerts these effects, although the process likely involves the ventral striatum given its function in motivational salience and appetitive reinforcement. In a double-blinded, placebo-controlled, crossover study, heavy drinking nondaily social smokers (ie, light smokers or 'chippers') were examined using functional magnetic resonance imaging after they ingested an acute dose of alcohol or placebo. We probed reactivity in the ventral striatum and other brain regions during exposure to visual smoking vs nonsmoking control cues. We found that alcohol enhanced self-reported ratings of desire to smoke, and in this context, significantly increased ventral striatum responses to smoking compared with control cues. In exploratory analyses, we observed that alcohol dampened orbitofrontal activity across both cue types, whereas dorsolateral prefrontal and anterior cingulate cortex activation to smoking cues was not affected by alcohol. This study bridges a pharmacological challenge approach to the study of brain reactivity to smoking cues, extends prior cigarette cue imaging studies to nondependent smokers, and elucidates a potential neurobiological mechanism to explain the co-consumption of alcohol and cigarettes in nondependent users.