Search

In generative modeling of neuroimaging data, such as dynamic causal modeling (DCM), one typically considers several alternative models, either to determine the most plausible explanation for observed data (Bayesian model selection) or to account for model uncertainty (Bayesian model averaging). Both procedures rest on estimates of the model evidence, a principled trade-off between model accuracy and complexity. In the context of DCM, the log evidence is usually approximated using variational Bayes. Although this approach is highly efficient, it makes distributional assumptions and is vulnerable to local extrema. This paper introduces the use of thermodynamic integration (TI) for Bayesian model selection and averaging in the context of DCM. TI is based on Markov chain Monte Carlo sampling which is asymptotically exact but orders of magnitude slower than variational Bayes. In this paper, we explain the theoretical foundations of TI, covering key concepts such as the free energy and its origins in statistical physics. Our aim is to convey an in-depth understanding of the method starting from its historical origin in statistical physics. In addition, we demonstrate the practical application of TI via a series of examples which serve to guide the user in applying this method. Furthermore, these examples demonstrate that, given an efficient implementation and hardware capable of parallel processing, the challenge of high computational demand can be overcome successfully. The TI implementation presented in this paper is freely available as part of the open source software TAPAS.

Regression dynamic causal modeling (rDCM) is a novel and computationally highly efficient method for inferring effective connectivity at the whole-brain level. While face and construct validity of rDCM have already been demonstrated, here we assessed its test-retest reliability-a test-theoretical property of particular importance for clinical applications-together with group-level consistency of connection-specific estimates and consistency of whole-brain connectivity patterns over sessions. Using the Human Connectome Project dataset for eight different paradigms (tasks and rest) and two different parcellation schemes, we found that rDCM provided highly consistent connectivity estimates at the group level across sessions. Second, while test-retest reliability was limited when averaging over all connections (range of mean intraclass correlation coefficient 0.24-0.42 over tasks), reliability increased with connection strength, with stronger connections showing good to excellent test-retest reliability. Third, whole-brain connectivity patterns by rDCM allowed for identifying individual participants with high (and in some cases perfect) accuracy. Comparing the test-retest reliability of rDCM connectivity estimates with measures of functional connectivity, rDCM performed favorably-particularly when focusing on strong connections. Generally, for all methods and metrics, task-based connectivity estimates showed greater reliability than those from the resting state. Our results underscore the potential of rDCM for human connectomics and clinical applications. Keywords: Connectomics; Effective connectivity; Generative model; Regression dynamic causal modeling; Test-retest reliability; rDCM

Psychiatry faces fundamental challenges with regard to mechanistically guided differential diagnosis, as well as prediction of clinical trajectories and treatment response of individual patients. This has motivated the genesis of two closely intertwined fields: (i) Translational Neuromodeling (TN), which develops “computational assays” for inferring patient-specific disease processes from neuroimaging, electrophysiological, and behavioral data; and (ii) Computational Psychiatry (CP), with the goal of incorporating computational assays into clinical decision making in everyday practice. In order to serve as objective and reliable tools for clinical routine, computational assays require end-to-end pipelines from raw data (input) to clinically useful information (output). While these are yet to be established in clinical practice, individual components of this general end-to-end pipeline are being developed and made openly available for community use. In this paper, we present the Translational Algorithms for Psychiatry-Advancing Science (TAPAS) software package, an open-source collection of building blocks for computational assays in psychiatry. Collectively, the tools in TAPAS presently cover several important aspects of the desired end-to-end pipeline, including: (i) tailored experimental designs and optimization of measurement strategy prior to data acquisition, (ii) quality control during data acquisition, and (iii) artifact correction, statistical inference, and clinical application after data acquisition. Here, we review the different tools within TAPAS and illustrate how these may help provide a deeper understanding of neural and cognitive mechanisms of disease, with the ultimate goal of establishing automatized pipelines for predictions about individual patients. We hope that the openly available tools in TAPAS will contribute to the further development of TN/CP and facilitate the translation of advances in computational neuroscience into clinically relevant c

Connectomics is essential for understanding large-scale brain networks but requires that individual connection estimates are neurobiologically interpretable. In particular, a principle of brain organization is that reciprocal connections between cortical areas are functionally asymmetric. This is a challenge for fMRI-based connectomics in humans where only undirected functional connectivity estimates are routinely available. By contrast, whole-brain estimates of effective (directed) connectivity are computationally challenging, and emerging methods require empirical validation. Here, using a motor task at 7T, we demonstrate that a novel generative model can infer known connectivity features in a whole-brain network (>200 regions, >40,000 connections) highly efficiently. Furthermore, graph-theoretical analyses of directed connectivity estimates identify functional roles of motor areas more accurately than undirected functional connectivity estimates. These results, which can be achieved in an entirely unsupervised manner, demonstrate the feasibility of inferring directed connections in whole-brain networks and open new avenues for human connectomics.

Drugs affecting neuromodulation, for example by dopamine or acetylcholine, take centre stage among therapeutic strategies in psychiatry. These neuromodulators can change both neuronal gain and synaptic plasticity and therefore affect electrophysiological measures. An important goal for clinical diagnostics is to exploit this effect in the reverse direction, i.e., to infer the status of specific neuromodulatory systems from electrophysiological measures. In this study, we provide proof-of-concept that the functional status of cholinergic (specifically muscarinic) receptors can be inferred from electrophysiological data using generative (dynamic causal) models. To this end, we used epidural EEG recordings over two auditory cortical regions during a mismatch negativity (MMN) paradigm in rats. All animals were treated, across sessions, with muscarinic receptor agonists and antagonists at different doses. Together with a placebo condition, this resulted in five levels of muscarinic receptor status. Using a dynamic causal model - embodying a small network of coupled cortical microcircuits - we estimated synaptic parameters and their change across pharmacological conditions. The ensuing parameter estimates associated with (the neuromodulation of) synaptic efficacy showed both graded muscarinic effects and predictive validity between agonistic and antagonistic pharmacological conditions. This finding illustrates the potential utility of generative models of electrophysiological data as computational assays of muscarinic function. In application to EEG data of patients from heterogeneous spectrum diseases, e.g. schizophrenia, such models might help identify subgroups of patients that respond differentially to cholinergic treatments.

Dynamic causal models (DCMs) of electrophysiological data allow, in principle, for inference on hidden, bulk synaptic function in neural circuits. The directed influences between the neuronal elements of modeled circuits are subject to delays due to the finite transmission speed of axonal connections. Ordinary differential equations are therefore not adequate to capture the ensuing circuit dynamics, and delay differential equations (DDEs) are required instead. Previous work has illustrated that the integration of DDEs in DCMs benefits from sophisticated integration schemes in order to ensure rigorous parameter estimation and correct model identification. However, integration schemes that have been proposed for DCMs either emphasize speed (at the possible expense of accuracy) or robustness (but with computational costs that are problematic in practice). In this technical note, we propose an alternative integration scheme that overcomes these shortcomings and offers high computational efficiency while correctly preserving the nature of delayed effects. This integration scheme is available as open-source code in the Translational Algorithms for Psychiatry-Advancing Science (TAPAS) toolbox and can be easily integrated into existing software (SPM) for the analysis of DCMs for electrophysiological data. While this paper focuses on its application to the convolution-based formalism of DCMs, the new integration scheme can be equally applied to more advanced formulations of DCMs (e.g. conductance based models). Our method provides a new option for electrophysiological DCMs that offers the speed required for scientific projects, but also the accuracy required for rigorous translational applications, e.g. in computational psychiatry.

"Resting-state" functional magnetic resonance imaging (rs-fMRI) is widely used to study brain connectivity. So far, researchers have been restricted to measures of functional connectivity that are computationally efficient but undirected, or to effective connectivity estimates that are directed but limited to small networks. Here, we show that a method recently developed for task-fMRI-regression dynamic causal modeling (rDCM)-extends to rs-fMRI and offers both directional estimates and scalability to whole-brain networks. First, simulations demonstrate that rDCM faithfully recovers parameter values over a wide range of signal-to-noise ratios and repetition times. Second, we test construct validity of rDCM in relation to an established model of effective connectivity, spectral DCM. Using rs-fMRI data from nearly 200 healthy participants, rDCM produces biologically plausible results consistent with estimates by spectral DCM. Importantly, rDCM is computationally highly efficient, reconstructing whole-brain networks (>200 areas) within minutes on standard hardware. This opens promising new avenues for connectomics.

Interoception, the perception of internal bodily states, is thought to be inextricably linked to affective qualities such as anxiety. Although interoception spans sensory to metacognitive processing, it is not clear whether anxiety is differentially related to these processing levels. Here we investigated this question in the domain of breathing, using computational modeling and high-field (7 T) fMRI to assess brain activity relating to dynamic changes in inspiratory resistance of varying predictability. Notably, the anterior insula was associated with both breathing-related prediction certainty and prediction errors, suggesting an important role in representing and updating models of the body. Individuals with low versus moderate anxiety traits showed differential anterior insula activity for prediction certainty. Multi-modal analyses of data from fMRI, computational assessments of breathing-related metacognition, and questionnaires demonstrated that anxiety-interoception links span all levels from perceptual sensitivity to metacognition, with strong effects seen at higher levels of interoceptive processes.

The mesiotemporal lobe (MTL) is implicated in many cognitive processes, is compromised in numerous brain disorders, and exhibits a gradual cytoarchitectural transition from six-layered parahippocampal isocortex to three-layered hippocampal allocortex. Leveraging an ultra-high-resolution histological reconstruction of a human brain, our study showed that the dominant axis of MTL cytoarchitectural differentiation follows the iso-to-allocortical transition and depth-specific variations in neuronal density. Projecting the histology-derived MTL model to in-vivo functional MRI, we furthermore determined how its cytoarchitecture underpins its intrinsic effective connectivity and association to large-scale networks. Here, the cytoarchitectural gradient was found to underpin intrinsic effective connectivity of the MTL, but patterns differed along the anterior-posterior axis. Moreover, while the iso-to-allocortical gradient parametrically represented the multiple-demand relative to task-negative networks, anterior-posterior gradients represented transmodal versus unimodal networks. Our findings establish that the combination of micro- and macrostructural features allow the MTL to represent dominant motifs of whole-brain functional organisation.

Patients with major depressive disorder (MDD) show heterogeneous treatment response and highly variable clinical trajectories: while some patients experience swift recovery, others show relapsing-remitting or chronic courses. Predicting individual clinical trajectories at an early stage is a key challenge for psychiatry and might facilitate individually tailored interventions. So far, however, reliable predictors at the single-patient level are absent. Here, we evaluated the utility of a machine learning strategy – generative embedding (GE) – which combines interpretable generative models with discriminative classifiers. Specifically, we used functional magnetic resonance imaging (fMRI) data of emotional face perception in 85 MDD patients from the NEtherlands Study of Depression and Anxiety (NESDA) who had been followed up over two years and classified into three subgroups with distinct clinical trajectories. Combining a generative model of effective (directed) connectivity with support vector machines (SVMs), we could predict whether a given patient would experience chronic depression vs. fast remission with a balanced accuracy of 79%. Gradual improvement vs. fast remission could still be predicted above-chance, but less convincingly, with a balanced accuracy of 61%. Generative embedding outperformed classification based on conventional (descriptive) features, such as functional connectivity or local activation estimates, which were obtained from the same data and did not allow for above-chance classification accuracy. Furthermore, predictive performance of GE could be assigned to a specific network property: the trial-by-trial modulation of connections by emotional content. Given the limited sample size of our study, the present results are preliminary but may serve as proof-of-concept, illustrating the potential of GE for obtaining clinical predictions that are interpretable in terms of network mechanisms. Our findings suggest that abnormal dynamic changes of con

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.

The development of whole-brain models that can infer effective (directed) connection strengths from fMRI data represents a central challenge for computational neuroimaging. A recently introduced generative model of fMRI data, regression dynamic causal modeling (rDCM), moves towards this goal as it scales gracefully to very large networks. However, large-scale networks with thousands of connections are difficult to interpret; additionally, one typically lacks information (data points per free parameter) for precise estimation of all model parameters. This paper introduces sparsity constraints to the variational Bayesian framework of rDCM as a solution to these problems in the domain of task-based fMRI. This sparse rDCM approach enables highly efficient effective connectivity analyses in whole-brain networks and does not require a priori assumptions about the network's connectivity structure but prunes fully (all-to-all) connected networks as part of model inversion. Following the derivation of the variational Bayesian update equations for sparse rDCM, we use both simulated and empirical data to assess the face validity of the model. In particular, we show that it is feasible to infer effective connection strengths from fMRI data using a network with more than 100 regions and 10,000 connections. This demonstrates the feasibility of whole-brain inference on effective connectivity from fMRI data - in single subjects and with a run-time below 1 min when using parallelized code. We anticipate that sparse rDCM may find useful application in connectomics and clinical neuromodeling - for example, for phenotyping individual patients in terms of whole-brain network structure.

The development of brain imaging techniques, in particular functional magnetic resonance imaging (fMRI), made it possible to non-invasively study the hemispheric lateralization of cognitive brain functions in large cohorts. Comprehensive models of hemispheric lateralization are, however, still missing and should not only account for the hemispheric specialization of individual brain functions, but also for the interactions among different lateralized cognitive processes (e.g., language and visuospatial processing). This calls for robust and reliable paradigms to study hemispheric lateralization for various cognitive functions. While numerous reliable imaging paradigms have been developed for language, which represents the most prominent left-lateralized brain function, the reliability of imaging paradigms investigating typically right-lateralized brain functions, such as visuospatial processing, has received comparatively less attention. In the present study, we aimed to establish an fMRI paradigm that robustly and reliably identifies right-hemispheric activation evoked by visuospatial processing in individual subjects. In a first study, we therefore compared three frequently used paradigms for assessing visuospatial processing and evaluated their utility to robustly detect right-lateralized brain activity on a single-subject level. In a second study, we then assessed the test-retest reliability of the so-called Landmark task-the paradigm that yielded the most robust results in study 1. At the single-voxel level, we found poor reliability of the brain activation underlying visuospatial attention. This suggests that poor signal-to-noise ratios can become a limiting factor for test-retest reliability. This represents a common detriment of fMRI paradigms investigating visuospatial attention in general and therefore highlights the need for careful considerations of both the possibilities and limitations of the respective fMRI paradigm-in particular, when being intereste

The development of large-scale network models that infer the effective (directed) connectivity among neuronal populations from neuroimaging data represents a key challenge for computational neuroscience. Dynamic causal models (DCMs) of neuroimaging and electrophysiological data are frequently used for inferring effective connectivity but are presently restricted to small graphs (typically up to 10 regions) in order to keep model inversion computationally feasible. Here, we present a novel variant of DCM for functional magnetic resonance imaging (fMRI) data that is suited to assess effective connectivity in large (whole-brain) networks. The approach rests on translating a linear DCM into the frequency domain and reformulating it as a special case of Bayesian linear regression. This paper derives regression DCM (rDCM) in detail and presents a variational Bayesian inversion method that enables extremely fast inference and accelerates model inversion by several orders of magnitude compared to classical DCM. Using both simulated and empirical data, we demonstrate the face validity of rDCM under different settings of signal-to-noise ratio (SNR) and repetition time (TR) of fMRI data. In particular, we assess the potential utility of rDCM as a tool for whole-brain connectomics by challenging it to infer effective connection strengths in a simulated whole-brain network comprising 66 regions and 300 free parameters. Our results indicate that rDCM represents a computationally highly efficient approach with promising potential for inferring whole-brain connectivity from individual fMRI data.

While the right-hemispheric lateralization of the face perception network is well established, recent evidence suggests that handedness affects the cerebral lateralization of face processing at the hierarchical level of the fusiform face area (FFA). However, the neural mechanisms underlying differential hemispheric lateralization of face perception in right- and left-handers are largely unknown. Using dynamic causal modeling (DCM) for fMRI, we aimed to unravel the putative processes that mediate handedness-related differences by investigating the effective connectivity in the bilateral core face perception network. Our results reveal an enhanced recruitment of the left FFA in left-handers compared to right-handers, as evidenced by more pronounced face-specific modulatory influences on both intra- and interhemispheric connections. As structural and physiological correlates of handedness-related differences in face processing, right- and left-handers varied with regard to their gray matter volume in the left fusiform gyrus and their pupil responses to face stimuli. Overall, these results describe how handedness is related to the lateralization of the core face perception network, and point to different neural mechanisms underlying face processing in right- and left-handers. In a wider context, this demonstrates the entanglement of structurally and functionally remote brain networks, suggesting a broader underlying process regulating brain lateralization.