Computational neuroimaging strategies for single patient predictions

Hierarchical Bayesian inference for concurrent model fitting and comparison for group studies

Computational modeling plays an important role in modern neuroscience research. Much previous research has relied on statistical methods, separately, to address two problems that are actually interdependent. First, given a particular computational model, Bayesian hierarchical techniques have been used to estimate individual variation in parameters over a population of subjects, leveraging their population-level distributions. Second, candidate models are themselves compared, and individual variation in the expressed model estimated, according to the fits of the models to each subject. The interdependence between these two problems arises because the relevant population for estimating parameters of a model depends on which other subjects express the model. Here, we propose a hierarchical Bayesian inference (HBI) framework for concurrent model comparison, parameter estimation and inference at the population level, combining previous approaches. We show that this framework has important advantages for both parameter estimation and model comparison theoretically and experimentally. The parameters estimated by the HBI show smaller errors compared to other methods. Model comparison by HBI is robust against outliers and is not biased towards overly simplistic models. Furthermore, the fully Bayesian approach of our theory enables researchers to make inference on group-level parameters by performing HBI t-test.

Computational modeling of brain and behavior plays an important role in modern neuroscience research. By deconstructing mechanisms of behavior and quantifying parameters of interest, computational modeling helps researchers to study brain-behavior mechanisms. In neuroscience studies, a dataset includes a number of samples, and often the question of interest is to characterize parameters of interest in a population: Do patients with attention-deficit hyperactive disorders exhibit lower learning rate than the general population? Do cognitive enhancers, such as Ritalin, enhance parameters influencing decision making? The success of these efforts heavily depends on statistical methods making inference about validity and robustness of estimated parameters, as well as generalizability of computational models. In this work, we present a novel method, hierarchical Bayesian inference, for concurrent model comparison, parameter estimation and inference at the population level. We show, both theoretically and experimentally, that our approach has important advantages over previous methods. The proposed method has implications for computational modeling research in group studies across many areas of psychology, neuroscience, and psychiatry.

Pathophysiological and cognitive mechanisms of fatigue in multiple sclerosis

Fatigue is one of the most common symptoms in multiple sclerosis (MS), with a major impact on patients' quality of life. Currently, treatment proceeds by trial and error with limited success, probably due to the presence of multiple different underlying mechanisms. Recent neuroscientific advances offer the potential to develop tools for differentiating these mechanisms in individual patients and ultimately provide a principled basis for treatment selection. However, development of these tools for differential diagnosis will require guidance by pathophysiological and cognitive theories that propose mechanisms which can be assessed in individual patients. This article provides an overview of contemporary pathophysiological theories of fatigue in MS and discusses how the mechanisms they propose may become measurable with emerging technologies and thus lay a foundation for future personalised treatments.

Deep Brain Stimulation Programming 2.0: Future Perspectives for Target Identification and Adaptive Closed Loop Stimulation

Deep brain stimulation has developed into an established treatment for movement disorders and is being actively investigated for numerous other neurological as well as psychiatric disorders. An accurate electrode placement in the target area and the effective programming of DBS devices are considered the most important factors for the individual outcome. Recent research in humans highlights the relevance of widespread networks connected to specific DBS targets. Improving the targeting of anatomical and functional networks involved in the generation of pathological neural activity will improve the clinical DBS effect and limit side-effects. Here, we offer a comprehensive overview over the latest research on target structures and targeting strategies in DBS. In addition, we provide a detailed synopsis of novel technologies that will support DBS programming and parameter selection in the future, with a particular focus on closed-loop stimulation and associated biofeedback signals.

Sex differences in network controllability as a predictor of executive function in youth

Executive function is a quintessential human capacity that emerges late in development and displays different developmental trends in males and females. Sex differences in executive function in youth have been linked to vulnerability to psychopathology as well as to behaviors that impinge on health, wellbeing, and longevity. Yet, the neurobiological basis of these differences is not well understood, in part due to the spatiotemporal complexity inherent in patterns of brain network maturation supporting executive function. Here we test the hypothesis that sex differences in impulsivity in youth stem from sex differences in the controllability of structural brain networks as they rewire over development. Combining methods from network neuroscience and network control theory, we characterize the network control properties of structural brain networks estimated from diffusion imaging data acquired in males and females in a sample of 879 youth aged 8-22 years. We summarize the control properties of these networks by estimating average and modal controllability, two statistics that probe the ease with which brain areas can drive the network towards easy versus difficult-to-reach states. We find that females have higher modal controllability in frontal, parietal, and subcortical regions while males have higher average controllability in frontal and subcortical regions. Furthermore, controllability profiles in males are negatively related to the false positive rate on a continuous performance task, a common measure of impulsivity. Finally, we find associations between average controllability and individual differences in activation during an n-back working memory task. Taken together, our findings support the notion that sex differences in the controllability of structural brain networks can partially explain sex differences in executive function. Controllability of structural brain networks also predicts features of task-relevant activation, suggesting the potential for controllability to represent context-specific constraints on network state more generally.

Predictors of survival and progression in behavioural variant frontotemporal dementia

BACKGROUND AND PURPOSE

Predicting the course of behavioural variant frontotemporal dementia (bvFTD) remains a major clinical challenge. This study aimed to identify factors that predict survival and clinical progression in bvFTD.

METHODS

Consecutive patients with clinically probable bvFTD were prospectively followed up over an 8-year period. Baseline neuropsychological variables, presence of a known pathogenic frontotemporal dementia gene mutation and a systematic visual magnetic resonance imaging assessment at baseline were examined as candidate predictors using multivariate modelling.

RESULTS

After screening 121 cases, the study cohort consisted of 75 patients with probable bvFTD, with a mean age of 60.8 ± 8.5 years, followed up for a mean duration of 7.2 ± 3.5 years from symptom onset. Median survival time from disease onset was 10.8 years and median survival, prior to transition to nursing home, was 8.9 years. A total of 25 of the 75 patients died during the study follow-up period. Survival without dependence was predicted by shorter disease duration at presentation (hazard ratio, 0.49, P = 0.001), greater atrophy in the anterior cingulate cortex (hazard ratio, 1.75, P = 0.047), older age (hazard ratio, 1.07, P = 0.026) and a higher burden of behavioural symptoms (hazard ratio, 1.04, P = 0.015). In terms of disease progression, presence of a known pathogenic frontotemporal dementia mutation (β = 0.46, P < 0.001) was the strongest predictor of progression. Deficits in letter fluency (β = -0.43, P = 0.017) and greater atrophy in the motor cortex (β = 0.51, P = 0.03) were also associated with faster progression.

CONCLUSIONS

This study provides novel clinical predictors of survival and progression in bvFTD. Our findings are likely to have an impact on prognostication and care planning in this difficult disease.

Models of persecutory delusions: a mechanistic insight into the early stages of psychosis

Identifying robust markers for predicting the onset of psychosis has been a key challenge for early detection research. Persecutory delusions are core symptoms of psychosis, and social cognition is particularly impaired in first-episode psychosis patients and individuals at risk for developing psychosis. Here, we propose new avenues for translation provided by hierarchical Bayesian models of behaviour and neuroimaging data applied in the context of social learning to target persecutory delusions. As it comprises a mechanistic model embedded in neurophysiology, the findings of this approach may shed light onto inference and neurobiological causes of transition to psychosis.

Connectomics of bipolar disorder: a critical review, and evidence for dynamic instabilities within interoceptive networks

The notion that specific cognitive and emotional processes arise from functionally distinct brain regions has lately shifted toward a connectivity-based approach that emphasizes the role of network-mediated integration across regions. The clinical neurosciences have likewise shifted from a predominantly lesion-based approach to a connectomic paradigm-framing disorders as diverse as stroke, schizophrenia (SCZ), and dementia as "dysconnection syndromes". Here we position bipolar disorder (BD) within this paradigm. We first summarise the disruptions in structural, functional and effective connectivity that have been documented in BD. Not surprisingly, these disturbances show a preferential impact on circuits that support emotional processes, cognitive control and executive functions. Those at high risk (HR) for BD also show patterns of connectivity that differ from both matched control populations and those with BD, and which may thus speak to neurobiological markers of both risk and resilience. We highlight research fields that aim to link brain network disturbances to the phenotype of BD, including the study of large-scale brain dynamics, the principles of network stability and control, and the study of interoception (the perception of physiological states). Together, these findings suggest that the affective dysregulation of BD arises from dynamic instabilities in interoceptive circuits which subsequently impact on fear circuitry and cognitive control systems. We describe the resulting disturbance as a "psychosis of interoception".

Personalized pathology maps to quantify diffuse and focal brain damage

Background and objectives

Quantitative MRI (qMRI) permits the quantification of brain changes compatible with inflammation, degeneration and repair in multiple sclerosis (MS) patients. In this study, we propose a new method to provide personalized maps of tissue alterations and longitudinal brain changes based on different qMRI metrics, which provide complementary information about brain pathology.

Methods

We performed baseline and two-years follow-up on (i) 13 relapsing-remitting MS patients and (ii) four healthy controls. A group consisting of up to 65 healthy controls was used to compute the reference distribution of qMRI metrics in healthy tissue. All subjects underwent 3T MRI examinations including T1, T2, T2* relaxation and Magnetization Transfer Ratio (MTR) imaging. We used a recent partial volume estimation algorithm to estimate the concentration of different brain tissue types on T1 maps; then, we computed a deviation map (z-score map) for each contrast at both time-points. Finally, we subtracted those deviation maps only for voxels showing a significant difference with healthy tissue in one of the time points, to obtain a difference map for each subject.

Results and conclusion

Control subjects did not show any significant z-score deviations or longitudinal z-score changes. On the other hand, MS patients showed brain regions with cross-sectional and longitudinal concomitant increase in T1, T2, T2* z-scores and decrease of MTR z-scores, suggesting brain tissue degeneration/loss. In the lesion periphery, we observed areas with cross-sectional and longitudinal decreased T1/T2 and slight decrease in T2* most likely related to iron accumulation. Moreover, we measured longitudinal decrease in T1, T2 - and to a lesser extent in T2* - as well as a concomitant increase in MTR, suggesting remyelination/repair.

In summary, we have developed a method that provides whole-brain personalized maps of cross-sectional and longitudinal changes in MS patients, which are computed in patient space. These maps may open new perspectives to complement and support radiological evaluation of brain damage for a given patient.

Computational Phenotyping: Using Models to Understand Individual Differences in Personality, Development, and Mental Illness

This paper reviews progress in the application of computational models to personality, developmental, and clinical neuroscience. We first describe the concept of a computational phenotype, a collection of parameters derived from computational models fit to behavioral and neural data. This approach represents individuals as points in a continuous parameter space, complementing traditional trait and symptom measures. One key advantage of this representation is that it is mechanistic: The parameters have interpretations in terms of cognitive processes, which can be translated into quantitative predictions about future behavior and brain activity. We illustrate with several examples how this approach has led to new scientific insights into individual differences, developmental trajectories, and psychopathology. We then survey some of the challenges that lay ahead.

Recent developments in the genetics of attention-deficit hyperactivity disorder

Attention-deficit hyperactivity disorder (ADHD) is a developmental psychiatric disorder that affects children and adults. ADHD is one of the psychiatric disorders with the strongest genetic basis according to familial, twin, and single nucleotide polymorphisms (SNP)-based epidemiological studies. In this review, we provide an update of recent insights into the genetic basis of ADHD. We discuss recent progress from genome-wide association studies (GWAS) looking at common variants as well as rare copy number variations. New analysis of gene groups, so-called functional ontologies, provide some insight into the gene networks afflicted, pointing to the role of neurodevelopmentally expressed gene networks. Bioinformatic methods, such as functional enrichment analysis and protein-protein network analysis, are used to highlight biological processes of likely relevance to the etiology of ADHD. Additionally, copy number variations seem to map on important pathways implicated in synaptic signaling and neurodevelopment. While some candidate gene associations of, for example, neurotransmitter receptors and signaling, have been replicated, they do not seem to explain significant variance in recent GWAS. We discuss insights from recent case-control SNP-GWAS that have presented the first whole-genome significant SNP in ADHD.

Biophysical Modeling of Large-Scale Brain Dynamics and Applications for Computational Psychiatry

Noninvasive neuroimaging has revolutionized the study of the organization of the human brain and how its structure and function are altered in psychiatric disorders. A critical explanatory gap lies in our mechanistic understanding of how systems-level neuroimaging biomarkers emerge from underlying synaptic-level perturbations associated with a disease state. We describe an emerging computational psychiatry approach leveraging biophysically based computational models of large-scale brain dynamics and their potential integration with clinical and pharmacological neuroimaging. In particular, we focus on neural circuit models, which describe how patterns of functional connectivity observed in resting-state functional magnetic resonance imaging emerge from neural dynamics shaped by inter-areal interactions through underlying structural connectivity defining long-range projections. We highlight the importance of local circuit physiological dynamics, in combination with structural connectivity, in shaping the emergent functional connectivity. Furthermore, heterogeneity of local circuit properties across brain areas, which impacts large-scale dynamics, may be critical for modeling whole-brain phenomena and alterations in psychiatric disorders and pharmacological manipulation. Finally, we discuss important directions for future model development and biophysical extensions, which will expand their utility to link clinical neuroimaging to neurobiological mechanisms.

Variability and reliability of effective connectivity within the core default mode network: A multi-site longitudinal spectral DCM study

Dynamic causal modelling (DCM) for resting state fMRI – namely spectral DCM – is a recently developed and widely adopted method for inferring effective connectivity in intrinsic brain networks. Most applications of spectral DCM have focused on group-averaged connectivity within large-scale intrinsic brain networks; however, the consistency of subject- and session-specific estimates of effective connectivity has not been evaluated. Establishing reliability (within subjects) is crucial for its clinical use; e.g., as a neurophysiological phenotype of disease progression. Effective connectivity during rest is likely to vary due to changes in cognitive, and physiological states. Quantifying these variations may help understand functional brain architectures – and inform clinical applications. In the present study, we investigated the consistency of effective connectivity within and between subjects, as well as potential sources of variability (e.g., hemispheric asymmetry). We also addressed the effects on consistency of standard data processing procedures. DCM analyses were applied to four longitudinal resting state fMRI datasets. Our sample comprised 17 subjects with 589 resting state fMRI sessions in total. These data allowed us to quantify the robustness of connectivity estimates for each subject, and to generalise our conclusions beyond specific data features. We found that subjects showed systematic and reliable patterns of hemispheric asymmetry. When asymmetry was taken into account, subjects showed very similar connectivity patterns. We also found that various processing procedures (e.g. global signal regression and ROI size) had little effect on inference and the reliability of connectivity estimates for the majority of subjects. Finally, Bayesian model reduction significantly increased the consistency of connectivity patterns.

•

Across datasets, subjects’ effective connectivity patterns in the core default mode network showed hemispheric asymmetry.

•

Differences in hemispheric asymmetry was found to be a major source of between-subject variability.

•

In contrast, most subjects showed reliable within-subject hemispheric asymmetry.

•

Differences in preprocessing methods had little effect on connectivity estimates.

•

Bayesian model reduction increased the within- and between-subject consistency of connectivity patterns.

Classification and Prediction of Brain Disorders Using Functional Connectivity: Promising but Challenging

Brain functional imaging data, especially functional magnetic resonance imaging (fMRI) data, have been employed to reflect functional integration of the brain. Alteration in brain functional connectivity (FC) is expected to provide potential biomarkers for classifying or predicting brain disorders. In this paper, we present a comprehensive review in order to provide guidance about the available brain FC measures and typical classification strategies. We survey the state-of-the-art FC analysis methods including widely used static functional connectivity (SFC) and more recently proposed dynamic functional connectivity (DFC). Temporal correlations among regions of interest (ROIs), data-driven spatial network and functional network connectivity (FNC) are often computed to reflect SFC from different angles. SFC can be extended to DFC using a sliding-window framework, and intrinsic connectivity states along the time-varying connectivity patterns are typically extracted using clustering or decomposition approaches. We also briefly summarize window-less DFC approaches. Subsequently, we highlight various strategies for feature selection including the filter, wrapper and embedded methods. In terms of model building, we include traditional classifiers as well as more recently applied deep learning methods. Moreover, we review representative applications with remarkable classification accuracy for psychosis and mood disorders, neurodevelopmental disorder, and neurological disorders using fMRI data. Schizophrenia, bipolar disorder, autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), Alzheimer's disease and mild cognitive impairment (MCI) are discussed. Finally, challenges in the field are pointed out with respect to the inaccurate diagnosis labeling, the abundant number of possible features and the difficulty in validation. Some suggestions for future work are also provided.

Variational Bayesian inversion for hierarchical unsupervised generative embedding (HUGE)

A recently introduced hierarchical generative model unified the inference of effective connectivity in individual subjects and the unsupervised identification of subgroups defined by connectivity patterns. This hierarchical unsupervised generative embedding (HUGE) approach combined a hierarchical formulation of dynamic causal modelling (DCM) for fMRI with Gaussian mixture models and relied on Markov chain Monte Carlo (MCMC) sampling for inference. While well suited for the inversion of complex hierarchical models, MCMC-based sampling suffers from a computational burden that is prohibitive for many applications. To address this problem, this paper derives an efficient variational Bayesian (VB) inversion scheme for HUGE that simultaneously provides approximations to the posterior distribution over model parameters and to the log model evidence. The face validity of the VB scheme was tested using two synthetic fMRI datasets with known ground truth. Additionally, an empirical fMRI dataset of stroke patients and healthy controls was used to evaluate the practical utility of the method in application to real-world problems. Our analyses demonstrate good performance of our VB scheme, with a marked speed-up of model inversion by two orders of magnitude compared to MCMC, while maintaining a similar level of accuracy. Notably, additional acceleration would be possible if parallel computing techniques were applied. Generally, our VB implementation of HUGE is fast enough to support multi-start procedures for whole-group analyses, a useful strategy to ameliorate problems with local extrema. HUGE thus represents a potentially useful practical solution for an important problem in clinical neuromodeling and computational psychiatry, i.e., the unsupervised detection of subgroups in heterogeneous populations that are defined by effective connectivity.

Improving the Standard for Deep Brain Stimulation Therapy: Target Structures and Feedback Signals for Adaptive Stimulation. Current Perspectives and Future Directions

Deep brain stimulation (DBS) is an established therapeutic option for the treatment of various neurological disorders and has been used successfully in movement disorders for over 25 years. However, the standard stimulation schemes have not changed substantially. Two major points of interest for the further development of DBS are target-structures and novel adaptive stimulation techniques integrating feedback signals. We describe recent research results on target structures and on neural and behavioural feedback signals for adaptive deep brain stimulation (aDBS), as well as outline future directions.

Interpreting and Utilising Intersubject Variability in Brain Function

We consider between-subject variance in brain function as data rather than noise. We describe variability as a natural output of a noisy plastic system (the brain) where each subject embodies a particular parameterisation of that system. In this context, variability becomes an opportunity to: (i) better characterise typical versus atypical brain functions; (ii) reveal the different cognitive strategies and processing networks that can sustain similar tasks; and (iii) predict recovery capacity after brain damage by taking into account both damaged and spared processing pathways. This has many ramifications for understanding individual learning preferences and explaining the wide differences in human abilities and disabilities. Understanding variability boosts the translational potential of neuroimaging findings, in particular in clinical and educational neuroscience.

Progressive neurodegeneration following spinal cord injury: Implications for clinical trials

Objective

To quantify atrophy, demyelination, and iron accumulation over 2 years following acute spinal cord injury and to identify MRI predictors of clinical outcomes and determine their suitability as surrogate markers of therapeutic intervention.

Methods

We assessed 156 quantitative MRI datasets from 15 patients with spinal cord injury and 18 controls at baseline and 2, 6, 12, and 24 months after injury. Clinical recovery (including neuropathic pain) was assessed at each time point. Between-group differences in linear and nonlinear trajectories of volume, myelin, and iron change were estimated. Structural changes by 6 months were used to predict clinical outcomes at 2 years.

Results

The majority of patients showed clinical improvement with recovery stabilizing at 2 years. Cord atrophy decelerated, while cortical white and gray matter atrophy progressed over 2 years. Myelin content in the spinal cord and cortex decreased progressively over time, while cerebellar loss decreases decelerated. As atrophy progressed in the thalamus, sustained iron accumulation was evident. Smaller cord and cranial corticospinal tract atrophy, and myelin changes within the sensorimotor cortices, by 6 months predicted recovery in lower extremity motor score at 2 years. Whereas greater cord atrophy and microstructural changes in the cerebellum, anterior cingulate cortex, and secondary sensory cortex by 6 months predicted worse sensory impairment and greater neuropathic pain intensity at 2 years.

Conclusion

These results draw attention to trauma-induced neuroplastic processes and highlight the intimate relationships among neurodegenerative processes in the cord and brain. These measurable changes are sufficiently large, systematic, and predictive to render them viable outcome measures for clinical trials.

Symptom dimensions in obsessive-compulsive disorder as predictors of neurobiology and treatment response

Purpose of review

Specific symptom dimensions of obsessive-compulsive disorder (OCD) have been suggested as an approach to reduce the heterogeneity of obsessive-compulsive disorder, predict treatment outcome, and relate to brain structure and function. Here, we review studies addressing these issues.

Recent findings

The contamination and symmetry/ordering dimensions have not been reliably associated with treatment outcome. Some studies found that greater severity of sexual/aggressive/religious symptoms predicted a worse outcome after cognitive behavioral therapy (CBT) and a better outcome after serotonin reuptake inhibitors (SRIs). Contamination symptoms have been related to increased amygdala and insula activation in a few studies, while sexual/aggressive/religious symptoms have also been related to more pronounced alterations in the function and structure of the amygdala. Increased pre-treatment limbic responsiveness has been related to better outcomes of CBT, but most imaging studies show important limitations and replication in large-scale studies is needed. We review possible reasons for the strong limbic involvement of the amygdala in patients with more sexual/aggressive/religious symptoms, in relation to their sensitivity to CBT.

Summary

Symptom dimensions may predict treatment outcome, and patients with sexual/religious/aggressive symptoms are at a greater risk of not starting or delaying treatment. This is likely partly due to more shame and perceived immorality which is also related to stronger amygdala response. Competently delivered CBT is likely to help these patients improve to the same degree as patients with other symptoms.

The relationship between spatial configuration and functional connectivity of brain regions

Brain connectivity is often considered in terms of the communication between functionally distinct brain regions. Many studies have investigated the extent to which patterns of coupling strength between multiple neural populations relates to behaviour. For example, studies have used ‘functional connectivity fingerprints’ to characterise individuals' brain activity. Here, we investigate the extent to which the exact spatial arrangement of cortical regions interacts with measures of brain connectivity. We find that the shape and exact location of brain regions interact strongly with the modelling of brain connectivity, and present evidence that the spatial arrangement of functional regions is strongly predictive of non-imaging measures of behaviour and lifestyle. We believe that, in many cases, cross-subject variations in the spatial configuration of functional brain regions are being interpreted as changes in functional connectivity. Therefore, a better understanding of these effects is important when interpreting the relationship between functional imaging data and cognitive traits.

People differ a lot from one another in terms of their personality, behaviour and lifestyle. This individuality is attributed to the different regions in the brain, and the strength of communication between them. The connectivity pattern between these areas is thought to be as unique as a fingerprint. If the connections are weak or disrupted it can play a role in conditions such as schizophrenia, depression or Alzheimer’s disease. It is thought that the strength of the connection depends on how strongly the nerve cells in these regions communicate. But are these individual differences solely caused by different strengths of connection, or could other factors contribute to them?

Now, Bijsterbosch et al. found that the size, shape and exact position of the brain regions was also strongly linked to the different behaviours of individuals. The study used brain scans, behavioural tests and questionnaires from a large database about lifestyle choices and demographics, to analyse the relationship between the different brain features of healthy individuals. The results showed that the variations in the brain regions were linked to many behavioural factors including intelligence, life satisfaction, drug use and aggression problems. Moreover, Bijsterbosch et al. showed that the existing methods for estimating the strength of connection between brain regions could reveal more about the spatial layout of these regions than the actual connection strength between them. This suggests that new approaches are needed to properly evaluate the strength of the connections.

Some psychiatric and neurological diseases may be associated with changes in size and position of the different regions in the brain. In future, the findings of this study could be applied to individuals affected by such conditions, to see if the location of a region could be used as a diagnostic indicator.

Generative models for clinical applications in computational psychiatry

Despite the success of modern neuroimaging techniques in furthering our understanding of cognitive and pathophysiological processes, translation of these advances into clinically relevant tools has been virtually absent until now. Neuromodeling represents a powerful framework for overcoming this translational deadlock, and the development of computational models to solve clinical problems has become a major scientific goal over the last decade, as reflected by the emergence of clinically oriented neuromodeling fields like Computational Psychiatry, Computational Neurology, and Computational Psychosomatics. Generative models of brain physiology and connectivity in the human brain play a key role in this endeavor, striving for computational assays that can be applied to neuroimaging data from individual patients for differential diagnosis and treatment prediction. In this review, we focus on dynamic causal modeling (DCM) and its use for Computational Psychiatry. DCM is a widely used generative modeling framework for functional magnetic resonance imaging (fMRI) and magneto-/electroencephalography (M/EEG) data. This article reviews the basic concepts of DCM, revisits examples where it has proven valuable for addressing clinically relevant questions, and critically discusses methodological challenges and recent methodological advances. We conclude this review with a more general discussion of the promises and pitfalls of generative models in Computational Psychiatry and highlight the path that lies ahead of us. This article is categorized under: Neuroscience > Computation Neuroscience > Clinical Neuroscience.

Stable Sparse Classifiers Identify qEEG Signatures that Predict Learning Disabilities (NOS) Severity

In this paper, we present a novel methodology to solve the classification problem, based on sparse (data-driven) regressions, combined with techniques for ensuring stability, especially useful for high-dimensional datasets and small samples number. The sensitivity and specificity of the classifiers are assessed by a stable ROC procedure, which uses a non-parametric algorithm for estimating the area under the ROC curve. This method allows assessing the performance of the classification by the ROC technique, when more than two groups are involved in the classification problem, i.e., when the gold standard is not binary. We apply this methodology to the EEG spectral signatures to find biomarkers that allow discriminating between (and predicting pertinence to) different subgroups of children diagnosed as Not Otherwise Specified Learning Disabilities (LD-NOS) disorder. Children with LD-NOS have notable learning difficulties, which affect education but are not able to be put into some specific category as reading (Dyslexia), Mathematics (Dyscalculia), or Writing (Dysgraphia). By using the EEG spectra, we aim to identify EEG patterns that may be related to specific learning disabilities in an individual case. This could be useful to develop subject-based methods of therapy, based on information provided by the EEG. Here we study 85 LD-NOS children, divided in three subgroups previously selected by a clustering technique over the scores of cognitive tests. The classification equation produced stable marginal areas under the ROC of 0.71 for discrimination between Group 1 vs. Group 2; 0.91 for Group 1 vs. Group 3; and 0.75 for Group 2 vs. Group1. A discussion of the EEG characteristics of each group related to the cognitive scores is also presented.

Data quality over data quantity in computational cognitive neuroscience

We analyzed factors that may hamper the advancement of computational cognitive neuroscience (CCN). These factors include a particular statistical mindset, which paves the way for the dominance of statistical power theory and a preoccupation with statistical replicability in the behavioral and neural sciences. Exclusive statistical concerns about sampling error occur at the cost of an inadequate representation of the problem of measurement error. We contrasted the manipulation of data quantity (sampling error, by varying the number of subjects) against the manipulation of data quality (measurement error, by varying the number of data per subject) in a simulated Bayesian model identifiability study. The results were clear-cut in showing that - across all levels of signal-to-noise ratios - varying the number of subjects was completely inconsequential, whereas the number of data per subject exerted massive effects on model identifiability. These results emphasize data quality over data quantity, and they call for the integration of statistics and measurement theory.

Machine Learning for Precision Psychiatry: Opportunities and Challenges

The nature of mental illness remains a conundrum. Traditional disease categories are increasingly suspected to misrepresent the causes underlying mental disturbance. Yet psychiatrists and investigators now have an unprecedented opportunity to benefit from complex patterns in brain, behavior, and genes using methods from machine learning (e.g., support vector machines, modern neural-network algorithms, cross-validation procedures). Combining these analysis techniques with a wealth of data from consortia and repositories has the potential to advance a biologically grounded redefinition of major psychiatric disorders. Increasing evidence suggests that data-derived subgroups of psychiatric patients can better predict treatment outcomes than DSM/ICD diagnoses can. In a new era of evidence-based psychiatry tailored to single patients, objectively measurable endophenotypes could allow for early disease detection, individualized treatment selection, and dosage adjustment to reduce the burden of disease. This primer aims to introduce clinicians and researchers to the opportunities and challenges in bringing machine intelligence into psychiatric practice.

Determining Excitatory and Inhibitory Neuronal Activity from Multimodal fMRI Data Using a Generative Hemodynamic Model

Hemodynamic responses, in general, and the blood oxygenation level-dependent (BOLD) fMRI signal, in particular, provide an indirect measure of neuronal activity. There is strong evidence that the BOLD response correlates well with post-synaptic changes, induced by changes in the excitatory and inhibitory (E-I) balance between active neuronal populations. Typical BOLD responses exhibit transients, such as the early-overshoot and post-stimulus undershoot, that can be linked to transients in neuronal activity, but they can also result from vascular uncoupling between cerebral blood flow (CBF) and venous cerebral blood volume (venous CBV). Recently, we have proposed a novel generative hemodynamic model of the BOLD signal within the dynamic causal modeling framework, inspired by physiological observations, called P-DCM (Havlicek et al., 2015). We demonstrated the generative model's ability to more accurately model commonly observed neuronal and vascular transients in single regions but also effective connectivity between multiple brain areas (Havlicek et al., 2017b). In this paper, we additionally demonstrate the versatility of the generative model to jointly explain dynamic relationships between neuronal and hemodynamic physiological variables underlying the BOLD signal using multi-modal data. For this purpose, we utilized three distinct data-sets of experimentally induced responses in the primary visual areas measured in human, cat, and monkey brain, respectively: (1) CBF and BOLD responses; (2) CBF, total CBV, and BOLD responses (Jin and Kim, 2008); and (3) positive and negative neuronal and BOLD responses (Shmuel et al., 2006). By fitting the generative model to the three multi-modal experimental data-sets, we showed that the presence or absence of dynamic features in the BOLD signal is not an unambiguous indication of presence or absence of those features on the neuronal level. Nevertheless, the generative model that takes into account the dynamics of the physiological mechanisms underlying the BOLD response allowed dissociating neuronal from vascular transients and deducing excitatory and inhibitory neuronal activity time-courses from BOLD data alone and from multi-modal data.

Laminar fMRI and computational theories of brain function

Recently developed methods for functional MRI at the resolution of cortical layers (laminar fMRI) offer a novel window into neurophysiological mechanisms of cortical activity. Beyond physiology, laminar fMRI also offers an unprecedented opportunity to test influential theories of brain function. Specifically, hierarchical Bayesian theories of brain function, such as predictive coding, assign specific computational roles to different cortical layers. Combined with computational models, laminar fMRI offers a unique opportunity to test these proposals noninvasively in humans. This review provides a brief overview of predictive coding and related hierarchical Bayesian theories, summarises their predictions with regard to layered cortical computations, examines how these predictions could be tested by laminar fMRI, and considers methodological challenges. We conclude by discussing the potential of laminar fMRI for clinically useful computational assays of layer-specific information processing.

Serial correlations in single-subject fMRI with sub-second TR

When performing statistical analysis of single-subject fMRI data, serial correlations need to be taken into account to allow for valid inference. Otherwise, the variability in the parameter estimates might be under-estimated resulting in increased false-positive rates. Serial correlations in fMRI data are commonly characterized in terms of a first-order autoregressive (AR) process and then removed via pre-whitening. The required noise model for the pre-whitening depends on a number of parameters, particularly the repetition time (TR). Here we investigate how the sub-second temporal resolution provided by simultaneous multislice (SMS) imaging changes the noise structure in fMRI time series. We fit a higher-order AR model and then estimate the optimal AR model order for a sequence with a TR of less than 600 ms providing whole brain coverage. We show that physiological noise modelling successfully reduces the required AR model order, but remaining serial correlations necessitate an advanced noise model. We conclude that commonly used noise models, such as the AR(1) model, are inadequate for modelling serial correlations in fMRI using sub-second TRs. Rather, physiological noise modelling in combination with advanced pre-whitening schemes enable valid inference in single-subject analysis using fast fMRI sequences.

Computational Psychosomatics and Computational Psychiatry: Toward a Joint Framework for Differential Diagnosis

This article outlines how a core concept from theories of homeostasis and cybernetics, the inference-control loop, may be used to guide differential diagnosis in computational psychiatry and computational psychosomatics. In particular, we discuss 1) how conceptualizing perception and action as inference-control loops yields a joint computational perspective on brain-world and brain-body interactions and 2) how the concrete formulation of this loop as a hierarchical Bayesian model points to key computational quantities that inform a taxonomy of potential disease mechanisms. We consider the utility of this perspective for differential diagnosis in concrete clinical applications.

Variational Bayesian Parameter Estimation Techniques for the General Linear Model

Variational Bayes (VB), variational maximum likelihood (VML), restricted maximum likelihood (ReML), and maximum likelihood (ML) are cornerstone parametric statistical estimation techniques in the analysis of functional neuroimaging data. However, the theoretical underpinnings of these model parameter estimation techniques are rarely covered in introductory statistical texts. Because of the widespread practical use of VB, VML, ReML, and ML in the neuroimaging community, we reasoned that a theoretical treatment of their relationships and their application in a basic modeling scenario may be helpful for both neuroimaging novices and practitioners alike. In this technical study, we thus revisit the conceptual and formal underpinnings of VB, VML, ReML, and ML and provide a detailed account of their mathematical relationships and implementational details. We further apply VB, VML, ReML, and ML to the general linear model (GLM) with non-spherical error covariance as commonly encountered in the first-level analysis of fMRI data. To this end, we explicitly derive the corresponding free energy objective functions and ensuing iterative algorithms. Finally, in the applied part of our study, we evaluate the parameter and model recovery properties of VB, VML, ReML, and ML, first in an exemplary setting and then in the analysis of experimental fMRI data acquired from a single participant under visual stimulation.

Neuroplasticity and the brain connectome: what can Jean Talairach’s reflections bring to modern psychosurgery?

Contrary to common psychosurgical practice in the 1950s, Dr. Jean Talairach had the intuition, based on clinical experience, that the brain connectome and neuroplasticity had a role to play in psychosurgery. Due to the remarkable progress of pharmacology at that time and to the technical limits of neurosurgery, these concepts were not put into practice. Currently, these concepts are being confirmed by modern techniques such as neuroimaging and computational neurosciences, and could pave the way for therapeutic innovation in psychiatry.

Psychosurgery commonly uses a localizationist approach, based on the idea that a lesion to a specific area is responsible for a deficit opposite to its function. To psychosurgeons such as Walter Freeman, who performed extensive lesions causing apparently inevitable deficit, Talairach answered with clinical data: complex psychic functions cannot be described that simply, because the same lesion does not provoke the same deficit in different patients. Moreover, cognitive impairment did not always follow efficacious psychosurgery. Talairach suggested that selectively destructing part of a network could open the door to a new organization, and that early psychotherapy could encourage this psychoplasticity. Talairach did not have the opportunity to put these concepts into practice in psychiatric diseases because of the sudden availability of neuroleptics, but connectomics and neuroplasticity gave rise to major advances in intraparenchymal neurosurgery, from epilepsy to low-grade glioma. In psychiatry, alongside long-standing theories implicating focal lesions and diffuse pathological processes, neuroimaging techniques are currently being developed. In mentally healthy individuals, combining diffusion tensor imaging with functional MRI, magnetoencephalography, and electroencephalography allows the determination of a comprehensive map of neural connections in the brain on many spatial scales, the so-called connectome. Ultimately, global neurocomputational models could predict physiological activity, behavior, and subjective feeling, and describe neuropsychiatric disorders.

Connectomic studies comparing psychiatric patients with controls have already confirmed the early intuitions of Talairach. As a striking example, massive dysconnectivity has been found in schizophrenia, leading some authors to propose a “dysconnection hypothesis.” Alterations of the connectome have also been demonstrated in obsessive-compulsive disorder and depression. Furthermore, normalization of the functional dysconnectivity has been observed following clinical improvement in several therapeutic interventions, from psychotherapy to pharmacological treatments. Provided that mental disorders result from abnormal structural or functional wiring, targeted psychosurgery would require that one be able: 1) to identify the pathological network involved in a given patient; 2) to use neurostimulation to safely create a reversible and durable alteration, mimicking a lesion, in a network compatible with neuroplasticity; and 3) to predict which functional lesion would result in adapted neuronal plasticity and/or to guide neuronal plasticity to promote recovery. All these conditions, already suggested by Talairach, could now be achievable considering modern biomarkers and surgical progress.

The Stochastic Early Reaction, Inhibition, and late Action (SERIA) model for antisaccades

The antisaccade task is a classic paradigm used to study the voluntary control of eye movements. It requires participants to suppress a reactive eye movement to a visual target and to concurrently initiate a saccade in the opposite direction. Although several models have been proposed to explain error rates and reaction times in this task, no formal model comparison has yet been performed. Here, we describe a Bayesian modeling approach to the antisaccade task that allows us to formally compare different models on the basis of their evidence. First, we provide a formal likelihood function of actions (pro- and antisaccades) and reaction times based on previously published models. Second, we introduce the Stochastic Early Reaction, Inhibition, and late Action model (SERIA), a novel model postulating two different mechanisms that interact in the antisaccade task: an early GO/NO-GO race decision process and a late GO/GO decision process. Third, we apply these models to a data set from an experiment with three mixed blocks of pro- and antisaccade trials. Bayesian model comparison demonstrates that the SERIA model explains the data better than competing models that do not incorporate a late decision process. Moreover, we show that the early decision process postulated by the SERIA model is, to a large extent, insensitive to the cue presented in a single trial. Finally, we use parameter estimates to demonstrate that changes in reaction time and error rate due to the probability of a trial type (pro- or antisaccade) are best explained by faster or slower inhibition and the probability of generating late voluntary prosaccades.

Evaluating the replicability, specificity, and generalizability of connectome fingerprints

Establishing reliable, robust, and unique brain signatures from neuroimaging data is a prerequisite for precision psychiatry, and therefore a highly sought-after goal in contemporary neuroscience. Recently, the procedure of connectome fingerprinting, using brain functional connectivity profiles as such signatures, was shown to be able to accurately identify individuals from a group of 126 subjects from the Human Connectome Project (HCP). However, the specificity and generalizability of this procedure were not tested. In this replication study, we show both for the original and an extended HCP data set (n = 900 subjects), as well as for an additional data set of more commonly acquired imaging quality (n = 84) that (i) although the high accuracy can be replicated for the larger HCP 900 data set, accuracy is (ii) lower for standard neuroimaging data, and, that (iii) connectome fingerprinting may not be specific enough to distinguish between individuals. In addition, both accuracy and specificity are projected to drop considerably as the size of a data set increases. Although the moderate-to-high accuracies do suggest there is a portion of unique variance, our results suggest that connectomes may actually be quite similar across individuals. This outcome may be relevant to how precision psychiatry could benefit from inferences based on functional connectomes.

Epigenetic mechanisms during ageing and neurogenesis as novel therapeutic avenues in human brain disorders

Ageing is the main risk factor for human neurological disorders. Among the diverse molecular pathways that govern ageing, epigenetics can guide age-associated decline in part by regulating gene expression and also through the modulation of genomic instability and high-order chromatin architecture. Epigenetic mechanisms are involved in the regulation of neural differentiation as well as in functional processes related to memory consolidation, learning or cognition during healthy lifespan. On the other side of the coin, many neurodegenerative diseases are associated with epigenetic dysregulation. The reversible nature of epigenetic factors and, especially, their role as mediators between the genome and the environment make them exciting candidates as therapeutic targets. Rather than providing a broad description of the pathways epigenetically deregulated in human neurological disorders, in this review, we have focused on the potential use of epigenetic enzymes as druggable targets to ameliorate neural decline during normal ageing and especially in neurological disorders. We will firstly discuss recent progress that supports a key role of epigenetic regulation during healthy ageing with an emphasis on the role of epigenetic regulation in adult neurogenesis. Then, we will focus on epigenetic alterations associated with ageing-related human disorders of the central nervous system. We will discuss examples in the context of psychiatric disorders, including schizophrenia and posttraumatic stress disorders, and also dementia or Alzheimer's disease as the most frequent neurodegenerative disease. Finally, methodological limitations and future perspectives are discussed.

Hyper-modulation of brain networks by the amygdala among women with Borderline Personality Disorder: Network signatures of affective interference during cognitive processing

Emotion dysregulation is a core characteristic of patients with Borderline Personality Disorder (BPD), and is often attributed to an imbalance in fronto-limbic network function. Hyperarousal of amygdala, especially in response to negative affective stimuli, results in affective interference with cognitive processing of executive functions. Clinical consequences include the impulsive-aggression, suicidal and self-injurious behaviors which characterize BPD. Dysfunctional interactions between amygdala and its network targets have not been well characterized during cognitive task performance. Using psychophysiological interaction analysis (PPI), we mapped network profiles of amygdala interaction with key regulatory regions during a Go No-Go task, modified to use negative, positive and neutral Ekman faces as targets. Fifty-six female subjects, 31 BPD and 25 healthy controls (HC), completed the affectively valenced Go No-Go task during fMRI scanning. In the negative affective condition, the amygdala exerted greater modulation of its targets in BPD compared to HC subjects in Rt. OFC, Rt. dACC, Rt. Parietal cortex, Rt. Basal Ganglia, and Rt. dlPFC. Across the spectrum of affective contrasts, hypermodulation in BPD subjects observed the following ordering: Negative > Neutral > Positive contrast. The amygdala seed exerted modulatory effects on specific target regions important in processing response inhibition and motor impulsiveness. The vulnerability of BPD subjects to affective interference with impulse control may be due to specific network dysfunction related to amygdala hyper-arousal and its effects on prefrontal regulatory regions such as the OFC and dACC.

Inference in the age of big data: Future perspectives on neuroscience

Neuroscience is undergoing faster changes than ever before. Over 100 years our field qualitatively described and invasively manipulated single or few organisms to gain anatomical, physiological, and pharmacological insights. In the last 10 years neuroscience spawned quantitative datasets of unprecedented breadth (e.g., microanatomy, synaptic connections, and optogenetic brain-behavior assays) and size (e.g., cognition, brain imaging, and genetics). While growing data availability and information granularity have been amply discussed, we direct attention to a less explored question: How will the unprecedented data richness shape data analysis practices? Statistical reasoning is becoming more important to distill neurobiological knowledge from healthy and pathological brain measurements. We argue that large-scale data analysis will use more statistical models that are non-parametric, generative, and mixing frequentist and Bayesian aspects, while supplementing classical hypothesis testing with out-of-sample predictions.