A deep learning approach for mental health quality prediction using functional network connectivity and assessment data.

While one can characterize mental health using questionnaires, such tools do not provide direct insight into the underlying biology. By linking approaches that visualize brain activity to questionnaires in the context of individualized prediction, we can gain new insights into the biology and behavioral aspects of brain health. Resting-state fMRI (rs-fMRI) can be used to identify biomarkers of these conditions and study patterns of abnormal connectivity. In this work, we estimate mental health quality for individual participants using static functional network connectivity (sFNC) data from rs-fMRI. The deep learning model uses the sFNC data as input to predict four categories of mental health quality and visualize the neural patterns indicative of each group. We used guided gradient class activation maps (guided Grad-CAM) to identify the most discriminative sFNC patterns. The effectiveness of this model was validated using the UK Biobank dataset, in which we showed that our approach outperformed four alternative models by 4-18% accuracy. The proposed model's performance evaluation yielded a classification accuracy of 76%, 78%, 88%, and 98% for the excellent, good, fair, and poor mental health categories, with poor mental health accuracy being the highest. The findings show distinct sFNC patterns across each group. The patterns associated with excellent mental health consist of the cerebellar-subcortical regions, whereas the most prominent areas in the poor mental health category are in the sensorimotor and visual domains. Thus the combination of rs-fMRI and deep learning opens a promising path for developing a comprehensive framework to evaluate and measure mental health. Moreover, this approach had the potential to guide the development of personalized interventions and enable the monitoring of treatment response. Overall this highlights the crucial role of advanced imaging modalities and deep learning algorithms in advancing our understanding and management of mental health.

A framework to quantify controlled directed interactions in network physiology applied to cognitive function assessment.

The complex nature of physiological systems where multiple organs interact to form a network is complicated by direct and indirect interactions, with varying strength and direction of influence. This study proposes a novel framework which quantifies directional and pairwise couplings, while controlling for the effect of indirect interactions. Simulation results confirm the superiority of this framework in uncovering directional primary links compared to previous published methods. In a practical application of cognitive attention and alertness tasks, the method was used to assess controlled directed interactions between the cardiac, respiratory and brain activities (prefrontal cortex). It revealed increased interactions during the alertness task between brain wave activity on the left side of the brain with heart rate and respiration compared to resting phases. During the attention task, an increased number of right brain wave interactions involving respiration was also observed compared to rest, in addition to left brain wave activity with heart rate. The proposed framework potentially assesses directional interactions in complex network physiology and may detect cognitive dysfunctions associated with altered network physiology.

Multimodal assessment shows misalignment of structural and functional thalamocortical connectivity in children and adolescents born very preterm.

Thalamocortical connections are altered following very preterm birth but it is unknown whether structural and functional alterations are linked and how they contribute to neurodevelopmental deficits. We used a multimodal approach in 27 very preterm and 35 term-born children and adolescents aged 10-16 years: Structural thalamocortical connectivity was quantified with two measures derived from probabilistic tractography of diffusion tensor data, namely the volume of thalamic segments with cortical connections and mean fractional anisotropy (FA) within the respective segments. High-density sleep EEG was recorded and sleep spindles were identified at each electrode. Sleep spindle density and integrated spindle activity (ISA) were calculated to quantify functional thalamocortical connectivity. In term-born participants, the volume of the global thalamic segment with cortical connections was strongly related to sleep spindles across the entire head (mean r â€‹= â€‹.53 â€‹± .10; range â€‹= â€‹0.35 to 0.78). Regionally, the volume of the thalamic segment connecting to frontal brain regions correlated with sleep spindle density in two clusters of electrodes over fronto-temporal brain regions (.42 â€‹± .06; 0.35 to 0.51 and 0.43 â€‹± .08; 0.35 to 0.62) and the volume of the thalamic segment connecting to parietal brain regions correlated with sleep spindle density over parietal brain regions (mean r â€‹= â€‹.43 â€‹± .07; 0.35 to 0.61). In very preterm participants, the volume of the thalamic segments was not associated with sleep spindles. In the very preterm group, mean FA within the global thalamic segment was negatively correlated with ISA over a cluster of frontal and temporo-occipital brain regions (mean r â€‹= â€‹-.53 â€‹± .07; -.41 to -.72). No association between mean FA and ISA was found in the term-born group. With this multimodal study protocol, we identified a potential misalignment between structural and functional thalamocortical connectivity in children and adolescents born very preterm. Eventually, this may shed further light on the neuronal mechanisms underlying neurodevelopmental sequelae of preterm birth.

Peer influence, Frontostriatal connectivity, and delay discounting in African American emerging adults.

Prior research has demonstrated the importance of delay discounting in adverse health behaviors, such as addiction, attention deficit hyperactivity disorder, risk taking, and obesity. Nevertheless, the functional connectivity of neural circuitry associated with delay discounting and the ways in which the social environment may influence frontostriatal connectivity remain largely unknown, particularly in African Americans. Building on recent literature implicating frontostriatal connectivity during active delay discounting decision making and at rest, we used functional magnetic resonance imaging to assess the association between delay discounting and frontostriatal resting state connectivity (rsFC). We also examined the capacity of social relationships with parents and peers to longitudinally predict frontostriatal rsFC. The study cohort was composed of 91 rural African American emerging adults followed over a 6-year period. Greater (i.e., more positive) frontostriatal rsFC was associated with decreased delay discounting (i.e., less impulsive decision making). In addition, peer relationships at ages 20 and 21 significantly predicted frontostriatal rsFC at age 25 above and beyond parental influence. A significant indirect effect of peer affiliation on delay discounting through frontostriatal rsFC also emerged. These results indicate a role of frontostriatal connectivity in delay discounting decision making and highlight peers' unique influence on decision making behaviors through frontostriatal rsFC during emerging adulthood.

Altered cross-talk between the hypothalamus and non-homeostatic regions linked to obesity and difficulty to lose weight.

Interactions between the hypothalamus and non-homeostatic regions may contribute to explain the difficulty to lose weight in obesity, an assumption never tested in human longitudinal studies. We investigated whether the functional connectivity between the medial and lateral hypothalamus (MH and LH) and corticostriatal regions differs between individuals with excess weight (n = 42) and normal weight (n = 39) using a seed-based resting-state approach. In addition, we examined the longitudinal association between functional connectivity and weight loss in a 3-month follow-up diet. Results showed that participants with excess weight had increased connectivity between the MH and the striatum and subgenual anterior cingulate cortex, and decreased connectivity with the middle frontal gyrus, and the bed nucleus of the stria terminalis (BNST), as well as a decreased connectivity between the LH and the cerebellum. Decreased connectivity between the MH and the posterior part of the BNST, and between the LH and the cerebellar cortex, predicted a greater percentage of weight loss. Functional connectivity measures explained 36% of the 3-month weight change among individuals with excess weight. We conclude that altered functional connectivity between homeostatic-hypothalamic regions and non-homeostatic corticostriatal and cerebellar regions is linked to obesity and difficulty to lose weight.

On the Role of Situational Stressors in the Disruption of Global Neural Network Stability during Problem Solving.

When individuals are placed in stressful situations, they are likely to exhibit deficits in cognitive capacity over and above situational demands. Despite this, individuals may still persevere and ultimately succeed in these situations. Little is known, however, about neural network properties that instantiate success or failure in both neutral and stressful situations, particularly with respect to regions integral for problem-solving processes that are necessary for optimal performance on more complex tasks. In this study, we outline how hidden Markov modeling based on multivoxel pattern analysis can be used to quantify unique brain states underlying complex network interactions that yield either successful or unsuccessful problem solving in more neutral or stressful situations. We provide evidence that brain network stability and states underlying synchronous interactions in regions integral for problem-solving processes are key predictors of whether individuals succeed or fail in stressful situations. Findings also suggested that individuals utilize discriminate neural patterns in successfully solving problems in stressful or neutral situations. Findings overall highlight how hidden Markov modeling can provide myriad possibilities for quantifying and better understanding the role of global network interactions in the problem-solving process and how the said interactions predict success or failure in different contexts.

Intrinsic connectivity of hippocampal subfields in normal elderly and mild cognitive impairment patients.

Hippocampal connectivity has been widely described but connectivity specificities of hippocampal subfields and their changes in early AD are poorly known. The aim of this study was to highlight hippocampal subfield networks in healthy elderly (HE) and their changes in amnestic patients with mild cognitive impairment (aMCI). Thirty-six HE and 27 aMCI patients underwent resting-state functional MRI scans. Specific intrinsic connectivity of bilateral CA1, SUB (subiculum), and CA2/3/4/DG was identified in HE (using seeds derived from manually delineation on high-resolution scans) and compared between HE and aMCI. Compared to the other subfields, CA1 was more strongly connected to the amygdala and occipital regions, CA2/3/4/DG to the left anterior cingulate cortex, temporal, and occipital regions, and SUB to the angular, precuneus, putamen, posterior cingulate, and frontal regions. aMCI patients showed reduced connectivity within the SUB network (with frontal and posterior cingulate regions). Our study highlighted for the first time three specific and distinct hippocampal subfield functional networks in HE, and their alterations in aMCI. These findings are important to understand AD specificities in both cognitive deficits and lesion topography, given the role of functional connectivity in these processes. Hum Brain Mapp 38:4922-4932, 2017. © 2017 Wiley Periodicals, Inc.

The evolution of cost-efficiency in neural networks during recovery from traumatic brain injury.

A somewhat perplexing finding in the systems neuroscience has been the observation that physical injury to neural systems may result in enhanced functional connectivity (i.e., hyperconnectivity) relative to the typical network response. The consequences of local or global enhancement of functional connectivity remain uncertain and this is particularly true for the overall metabolic cost of the network. We examine the hyperconnectivity hypothesis in a sample of 14 individuals with TBI with data collected at approximately 3, 6, and 12 months following moderate and severe TBI. As anticipated, individuals with TBI showed increased network strength and cost early after injury, but by one-year post injury hyperconnectivity was more circumscribed to frontal DMN and temporal-parietal attentional control regions. Cost in these subregions was a significant predictor of cognitive performance. Cost-efficiency analysis in the Power 264 data parcellation suggested that at 6 months post injury the network requires higher cost connections to achieve high efficiency as compared to the network 12 months post injury. These results demonstrate that networks self-organize to re-establish connectivity while balancing cost-efficiency trade-offs.

Auditory processing assessment suggests that Wistar audiogenic rat neural networks are prone to entrainment.

Epilepsy is a neurological disease related to the occurrence of pathological oscillatory activity, but the basic physiological mechanisms of seizure remain to be understood. Our working hypothesis is that specific sensory processing circuits may present abnormally enhanced predisposition for coordinated firing in the dysfunctional brain. Such facilitated entrainment could share a similar mechanistic process as those expediting the propagation of epileptiform activity throughout the brain. To test this hypothesis, we employed the Wistar audiogenic rat (WAR) reflex animal model, which is characterized by having seizures triggered reliably by sound. Sound stimulation was modulated in amplitude to produce an auditory steady-state-evoked response (ASSR; -53.71Hz) that covers bottom-up and top-down processing in a time scale compatible with the dynamics of the epileptic condition. Data from inferior colliculus (IC) c-Fos immunohistochemistry and electrographic recordings were gathered for both the control Wistar group and WARs. Under 85-dB SLP auditory stimulation, compared to controls, the WARs presented higher number of Fos-positive cells (at IC and auditory temporal lobe) and a significant increase in ASSR-normalized energy. Similarly, the 110-dB SLP sound stimulation also statistically increased ASSR-normalized energy during ictal and post-ictal periods. However, at the transition from the physiological to pathological state (pre-ictal period), the WAR ASSR analysis demonstrated a decline in normalized energy and a significant increase in circular variance values compared to that of controls. These results indicate an enhanced coordinated firing state for WARs, except immediately before seizure onset (suggesting pre-ictal neuronal desynchronization with external sensory drive). These results suggest a competing myriad of interferences among different networks that after seizure onset converge to a massive oscillatory circuit.

Network-level assessment of reward-related activation in patients with ADHD and healthy individuals.

INTRODUCTION: Reward processing is a key aspect of cognitive control processes, putatively instantiated by mesolimbic and mesocortical brain circuits. Deficient signaling within these circuits has been associated with psychopathology. We applied a network discovery approach to assess specific functional networks associated with reward processing in participants with attention-deficit/hyperactivity disorder (ADHD). METHODS: To describe task-related processes in terms of integrated functional networks, we applied independent component analysis (ICA) to task response maps of 60 healthy participants who performed a monetary incentive delay (MID) task. The resulting components were interpreted on the basis of their similarity with group-level task responses as well as their similarity with brain networks derived from resting state fMRI analyses. ADHD-related effects on network characteristics including functional connectivity and communication between networks were examined in an independent sample comprising 150 participants with ADHD and 48 healthy controls. RESULTS: We identified 23 components to be associated with 4 large-scale functional networks: the default-mode, visual, executive control, and salience networks. The salience network showed a specific association with reward processing as well as the highest degree of within-network integration. ADHD was associated with decreased functional connectivity between the salience and executive control networks as well as with peripheral brain regions. CONCLUSIONS: Reward processing as measured with the MID task involves one reward-specific and three general functional networks. Participants with ADHD exhibited alterations in connectivity of both the salience and executive control networks and associated brain regions during task performance. Hum Brain Mapp 38:2359-2369, 2017. © 2017 Wiley Periodicals, Inc.

Learning effective connectivity from fMRI using autoregressive hidden Markov model with missing data.

BACKGROUND: Effective connectivity (EC) analysis of neuronal groups using fMRI delivers insights about functional-integration. However, fMRI signal has low-temporal resolution due to down-sampling and indirectly measures underlying neuronal activity. NEW METHOD: The aim is to address above issues for more reliable EC estimates. This paper proposes use of autoregressive hidden Markov model with missing data (AR-HMM-md) in dynamically multi-linked (DML) framework for learning EC using multiple fMRI time series. In our recent work (Dang et al., 2016), we have shown how AR-HMM-md for modelling single fMRI time series outperforms the existing methods. AR-HMM-md models unobserved neuronal activity and lost data over time as variables and estimates their values by joint optimization given fMRI observation sequence. RESULTS: The effectiveness in learning EC is shown using simulated experiments. Also the effects of sampling and noise are studied on EC. Moreover, classification-experiments are performed for Attention-Deficit/Hyperactivity Disorder subjects and age-matched controls for performance evaluation of real data. Using Bayesian model selection, we see that the proposed model converged to higher log-likelihood and demonstrated that group-classification can be performed with higher cross-validation accuracy of above 94% using distinctive network EC which characterizes patients vs. CONTROLS: The full data EC obtained from DML-AR-HMM-md is more consistent with previous literature than the classical multivariate Granger causality method. COMPARISON: The proposed architecture leads to reliable estimates of EC than the existing latent models. CONCLUSIONS: This framework overcomes the disadvantage of low-temporal resolution and improves cross-validation accuracy significantly due to presence of missing data variables and autoregressive process.

Defining the brain circuits involved in psychiatric disorders: IMI-NEWMEDS.

Despite the vast amount of research on schizophrenia and depression in the past two decades, there have been few innovative drugs to treat these disorders. Precompetitive research collaborations between companies and academic groups can help tackle this innovation deficit, as illustrated by the achievements of the IMI-NEWMEDS consortium.

Chronic Pain: Where the Body Meets the Brain.

Chronic musculoskeletal pain is one of the most intractable clinical problems faced by clinicians and can be devastating for patients. Central pain amplification is perceived pain that cannot be fully explained on the basis of somatic or neuropathic processes and is due to physiologic alterations in pain transmission or descending pain modulatory pathways. In any individual, central pain amplification may complicate nociceptive or neuropathic pain. Furthermore, patients with somatic symptom disorders may have alterations in their psychological or behavioral responses to pain that contribute significantly to the clinical presentation. Genetic, physiologic, and psychological factors associated with central pain amplification are beginning to be understood. One important contributor to chronic pain is perceived stress and stress response systems. We and others have shown a complex relationship between the physiologic stress response and chronic pain symptoms. Unfortunately, treatments for chronic pain are woefully inadequate and often worsen clinical outcomes. Developing new treatment strategies for patients with chronic pain is of utmost urgency. This essay provides a framework for thinking about chronic pain and developing new treatment approaches.

Scientometrics of drug discovery efforts: pain-related molecular targets.

The aim of this study was to make a scientometric assessment of drug discovery efforts centered on pain-related molecular targets. The following scientometric indices were used: the popularity index, representing the share of articles (or patents) on a specific topic among all articles (or patents) on pain over the same 5-year period; the index of change, representing the change in the number of articles (or patents) on a topic from one 5-year period to the next; the index of expectations, representing the ratio of the number of all types of articles on a topic in the top 20 journals relative to the number of articles in all (>5,000) biomedical journals covered by PubMed over a 5-year period; the total number of articles representing Phase I-III trials of investigational drugs over a 5-year period; and the trial balance index, a ratio of Phase I-II publications to Phase III publications. Articles (PubMed database) and patents (US Patent and Trademark Office database) on 17 topics related to pain mechanisms were assessed during six 5-year periods from 1984 to 2013. During the most recent 5-year period (2009-2013), seven of 17 topics have demonstrated high research activity (purinergic receptors, serotonin, transient receptor potential channels, cytokines, gamma aminobutyric acid, glutamate, and protein kinases). However, even with these seven topics, the index of expectations decreased or did not change compared with the 2004-2008 period. In addition, publications representing Phase I-III trials of investigational drugs (2009-2013) did not indicate great enthusiasm on the part of the pharmaceutical industry regarding drugs specifically designed for treatment of pain. A promising development related to the new tool of molecular targeting, ie, monoclonal antibodies, for pain treatment has not yet resulted in real success. This approach has not yet demonstrated clinical effectiveness (at least with nerve growth factor) much beyond conventional analgesics, when its potential cost is more than an order of magnitude higher than that of conventional treatments. This scientometric assessment demonstrated a lack of real breakthrough developments.

Stress assessment based on EEG univariate features and functional connectivity measures.

The biological response to stress originates in the brain but involves different biochemical and physiological effects. Many common clinical methods to assess stress are based on the presence of specific hormones and on features extracted from different signals, including electrocardiogram, blood pressure, skin temperature, or galvanic skin response. The aim of this paper was to assess stress using EEG-based variables obtained from univariate analysis and functional connectivity evaluation. Two different stressors, the Stroop test and sleep deprivation, were applied to 30 volunteers to find common EEG patterns related to stress effects. Results showed a decrease of the high alpha power (11 to 12 Hz), an increase in the high beta band (23 to 36 Hz, considered a busy brain indicator), and a decrease in the approximate entropy. Moreover, connectivity showed that the high beta coherence and the interhemispheric nonlinear couplings, measured by the cross mutual information function, increased significantly for both stressors, suggesting that useful stress indexes may be obtained from EEG-based features.

Anesthesia and neuroimaging: investigating the neural correlates of unconsciousness.

In the past 15 years, rapid technological development in the field of neuroimaging has led to a resurgence of interest in the study of consciousness. However, the neural bases of consciousness and the boundaries of unconscious processing remain poorly understood. Anesthesia combined with functional neuroimaging presents a unique approach for studying neural responses as a function of consciousness. In this review we summarize findings from functional neuroimaging studies that have used anesthetic drugs to study cognition at different levels of conscious awareness. We relate the results to those of psychophysical studies of cognition and explore their potential usefulness in interpreting clinical findings from studies of non-responsive patients.

Characterizing and Differentiating Brain State Dynamics via Hidden Markov Models.

Functional connectivity measured from resting state fMRI (R-fMRI) data has been widely used to examine the brain's functional activities and has been recently used to characterize and differentiate brain conditions. However, the dynamical transition patterns of the brain's functional states have been less explored. In this work, we propose a novel computational framework to quantitatively characterize the brain state dynamics via hidden Markov models (HMMs) learned from the observations of temporally dynamic functional connectomics, denoted as functional connectome states. The framework has been applied to the R-fMRI dataset including 44 post-traumatic stress disorder (PTSD) patients and 51 normal control (NC) subjects. Experimental results show that both PTSD and NC brains were undergoing remarkable changes in resting state and mainly transiting amongst a few brain states. Interestingly, further prediction with the best-matched HMM demonstrates that PTSD would enter into, but could not disengage from, a negative mood state. Importantly, 84% of PTSD patients and 86% of NC subjects are successfully classified via multiple HMMs using majority voting.

Assessment and modulation of resting-state neural networks after stroke.

PURPOSE OF REVIEW: Stroke is a major cause of disability; however, most patients experience spontaneous partial recovery of functions in subacute to chronic phases. Poststroke loss and recovery of functions have been increasingly correlated with brain-wide alterations in the connectivity of neural networks, which is described in this review. Elucidation of the mechanisms of functional brain remodeling could reveal targets and strategies for more effective neurorehabilitation. RECENT FINDINGS: Data from recent resting-state functional MRI, electroencephalography, magnetoencephalography, and optical imaging studies in patients and animal models have demonstrated that loss of function after stroke is closely associated with disrupted connectivity in large-scale networks beyond the lesion territory. Restoration of functional connectivity in the surviving networks appears to be critical for functional recovery, and this may be promoted with specific therapeutic strategies, such as robot-assisted training and noninvasive brain stimulation. The adaptability of functional networks relies on the structural integrity of neuronal pathways, but the relationship between the two remains incompletely understood. Furthermore, disturbed neurovascular coupling after stroke can confound hemodynamically based measurements of functional connectivity. SUMMARY: Identification of key network processes in adaptive brain plasticity can aid in the prediction of functional outcome and the development of therapeutic interventions to support and promote recovery after stroke.

Disrupted resting-state functional connectivity in minimally treated chronic schizophrenia.

OBJECTIVE: The pathophysiology of chronic schizophrenia may reflect long term brain changes related to the disorder. The effect of chronicity on intrinsic functional connectivity patterns in schizophrenia without the potentially confounding effect of antipsychotic medications, however, remains largely unknown. METHOD: We collected resting-state fMRI data in 21 minimally treated chronic schizophrenia patients and 20 healthy controls. We computed regional functional connectivity strength for each voxel in the brain, and further divided regional functional connectivity strength into short-range regional functional connectivity strength and long-range regional functional connectivity strength. General linear models were used to detect between-group differences in these regional functional connectivity strength metrics and to further systematically investigate the relationship between these differences and clinical/behavioral variables in the patients. RESULTS: Compared to healthy controls, the minimally treated chronic schizophrenia patients showed an overall reduced regional functional connectivity strength especially in bilateral sensorimotor cortex, right lateral prefrontal cortex, left insula and right lingual gyrus, and these regional functional connectivity strength decreases mainly resulted from disruption of short-range regional functional connectivity strength. The minimally treated chronic schizophrenia patients also showed reduced long-range regional functional connectivity strength in the bilateral posterior cingulate cortex/precuneus, and increased long-range regional functional connectivity strength in the right lateral prefrontal cortex and lingual gyrus. Notably, disrupted short-range regional functional connectivity strength mainly correlated with duration of illness and negative symptoms, whereas disrupted long-range regional functional connectivity strength correlated with neurocognitive performance. All of the results were corrected using Monte-Carlo simulation. CONCLUSIONS: This exploratory study demonstrates a disruption of intrinsic functional connectivity without long-term exposure to antipsychotic medications in chronic schizophrenia. Furthermore, this disruption was connection-distance dependent, thus raising the possibility for differential neural pathways in neurocognitive impairment and psychiatric symptoms in schizophrenia.

Barriers to investigator-initiated deep brain stimulation and device research.

The success of device-based research in the clinical neurosciences has overshadowed a critical and emerging problem in the biomedical research environment in the United States. Neuroprosthetic devices, such as deep brain stimulation (DBS), have been shown in humans to be promising technologies for scientific exploration of neural pathways and as powerful treatments. Large device companies have, over the past several decades, funded and developed major research programs. However, both the structure of clinical trial funding and the current regulation of device research threaten investigator-initiated efforts in neurologic disorders. The current atmosphere dissuades clinical investigators from pursuing formal and prospective research with novel devices or novel indications. We review our experience in conducting a federally funded, investigator-initiated, device-based clinical trial that utilized DBS for thalamic pain syndrome. We also explore barriers that clinical investigators face in conducting device-based clinical trials, particularly in early-stage studies or small disease populations. We discuss 5 specific areas for potential reform and integration: (1) alternative pathways for device approval; (2) eliminating right of reference requirements; (3) combining federal grant awards with regulatory approval; (4) consolidation of oversight for human subjects research; and (5) private insurance coverage for clinical trials. Careful reformulation of regulatory policy and funding mechanisms is critical for expanding investigator-initiated device research, which has great potential to benefit science, industry, and, most importantly, patients.