Amortized Bayesian inference on generative dynamical network models of epilepsy using deep neural density estimators.

Whole-brain modeling of epilepsy combines personalized anatomical data with dynamical models of abnormal activities to generate spatio-temporal seizure patterns as observed in brain imaging data. Such a parametric simulator is equipped with a stochastic generative process, which itself provides the basis for inference and prediction of the local and global brain dynamics affected by disorders. However, the calculation of likelihood function at whole-brain scale is often intractable. Thus, likelihood-free algorithms are required to efficiently estimate the parameters pertaining to the hypothetical areas, ideally including the uncertainty. In this study, we introduce the simulation-based inference for the virtual epileptic patient model (SBI-VEP), enabling us to amortize the approximate posterior of the generative process from a low-dimensional representation of whole-brain epileptic patterns. The state-of-the-art deep learning algorithms for conditional density estimation are used to readily retrieve the statistical relationships between parameters and observations through a sequence of invertible transformations. We show that the SBI-VEP is able to efficiently estimate the posterior distribution of parameters linked to the extent of the epileptogenic and propagation zones from sparse intracranial electroencephalography recordings. The presented Bayesian methodology can deal with non-linear latent dynamics and parameter degeneracy, paving the way for fast and reliable inference on brain disorders from neuroimaging modalities.

Delineating epileptogenic networks using brain imaging data and personalized modeling in drug-resistant epilepsy.

Precise estimates of epileptogenic zone networks (EZNs) are crucial for planning intervention strategies to treat drug-resistant focal epilepsy. Here, we present the virtual epileptic patient (VEP), a workflow that uses personalized brain models and machine learning methods to estimate EZNs and to aid surgical strategies. The structural scaffold of the patient-specific whole-brain network model is constructed from anatomical T1 and diffusion-weighted magnetic resonance imaging. Each network node is equipped with a mathematical dynamical model to simulate seizure activity. Bayesian inference methods sample and optimize key parameters of the personalized model using functional stereoelectroencephalography recordings of patients' seizures. These key parameters together with their personalized model determine a given patient's EZN. Personalized models were further used to predict the outcome of surgical intervention using virtual surgeries. We evaluated the VEP workflow retrospectively using 53 patients with drug-resistant focal epilepsy. VEPs reproduced the clinically defined EZNs with a precision of 0.6, where the physical distance between epileptogenic regions identified by VEP and the clinically defined EZNs was small. Compared with the resected brain regions of 25 patients who underwent surgery, VEP showed lower false discovery rates in seizure-free patients (mean, 0.028) than in non-seizure-free patients (mean, 0.407). VEP is now being evaluated in an ongoing clinical trial (EPINOV) with an expected 356 prospective patients with epilepsy.

Identifying spatio-temporal seizure propagation patterns in epilepsy using Bayesian inference.

Focal drug resistant epilepsy is a neurological disorder characterized by seizures caused by abnormal activity originating in one or more regions together called as epileptogenic zone. Treatment for such patients involves surgical resection of affected regions. Epileptogenic zone is typically identified using stereotactic EEG recordings from the electrodes implanted into the patient's brain. Identifying the epileptogenic zone is a challenging problem due to the spatial sparsity of electrode implantation. We propose a probabilistic hierarchical model of seizure propagation patterns, based on a phenomenological model of seizure dynamics called Epileptor. Using Bayesian inference, the Epileptor model is optimized to build patient specific virtual models that best fit to the log power of intracranial recordings. First, accuracy of the model predictions and identifiability of the model are investigated using synthetic data. Then, model predictions are evaluated against a retrospective patient cohort of 25 patients with varying surgical outcomes. In the patients who are seizure free after surgery, model predictions showed good match with the clinical hypothesis. In patients where surgery failed to achieve seizure freedom model predictions showed a strong mismatch. Our results demonstrate that proposed probabilistic model could be a valuable tool to aid the clinicians in identifying the seizure focus.

Multiscale dynamic mean field (MDMF) model relates resting-state brain dynamics with local cortical excitatory-inhibitory neurotransmitter homeostasis.

Previous computational models have related spontaneous resting-state brain activity with local excitatory-inhibitory balance in neuronal populations. However, how underlying neurotransmitter kinetics associated with E-I balance govern resting-state spontaneous brain dynamics remains unknown. Understanding the mechanisms by virtue of which fluctuations in neurotransmitter concentrations, a hallmark of a variety of clinical conditions, relate to functional brain activity is of critical importance. We propose a multiscale dynamic mean field (MDMF) model-a system of coupled differential equations for capturing the synaptic gating dynamics in excitatory and inhibitory neural populations as a function of neurotransmitter kinetics. Individual brain regions are modeled as population of MDMF and are connected by realistic connection topologies estimated from diffusion tensor imaging data. First, MDMF successfully predicts resting-state functional connectivity. Second, our results show that optimal range of glutamate and GABA neurotransmitter concentrations subserve as the dynamic working point of the brain, that is, the state of heightened metastability observed in empirical blood-oxygen-level-dependent signals. Third, for predictive validity the network measures of segregation (modularity and clustering coefficient) and integration (global efficiency and characteristic path length) from existing healthy and pathological brain network studies could be captured by simulated functional connectivity from an MDMF model.

On the influence of prior information evaluated by fully Bayesian criteria in a personalized whole-brain model of epilepsy spread.

Individualized anatomical information has been used as prior knowledge in Bayesian inference paradigms of whole-brain network models. However, the actual sensitivity to such personalized information in priors is still unknown. In this study, we introduce the use of fully Bayesian information criteria and leave-one-out cross-validation technique on the subject-specific information to assess different epileptogenicity hypotheses regarding the location of pathological brain areas based on a priori knowledge from dynamical system properties. The Bayesian Virtual Epileptic Patient (BVEP) model, which relies on the fusion of structural data of individuals, a generative model of epileptiform discharges, and a self-tuning Monte Carlo sampling algorithm, is used to infer the spatial map of epileptogenicity across different brain areas. Our results indicate that measuring the out-of-sample prediction accuracy of the BVEP model with informative priors enables reliable and efficient evaluation of potential hypotheses regarding the degree of epileptogenicity across different brain regions. In contrast, while using uninformative priors, the information criteria are unable to provide strong evidence about the epileptogenicity of brain areas. We also show that the fully Bayesian criteria correctly assess different hypotheses about both structural and functional components of whole-brain models that differ across individuals. The fully Bayesian information-theory based approach used in this study suggests a patient-specific strategy for epileptogenicity hypothesis testing in generative brain network models of epilepsy to improve surgical outcomes.

Data-driven method to infer the seizure propagation patterns in an epileptic brain from intracranial electroencephalography.

Surgical interventions in epileptic patients aimed at the removal of the epileptogenic zone have success rates at only 60-70%. This failure can be partly attributed to the insufficient spatial sampling by the implanted intracranial electrodes during the clinical evaluation, leading to an incomplete picture of spatio-temporal seizure organization in the regions that are not directly observed. Utilizing the partial observations of the seizure spreading through the brain network, complemented by the assumption that the epileptic seizures spread along the structural connections, we infer if and when are the unobserved regions recruited in the seizure. To this end we introduce a data-driven model of seizure recruitment and propagation across a weighted network, which we invert using the Bayesian inference framework. Using a leave-one-out cross-validation scheme on a cohort of 45 patients we demonstrate that the method can improve the predictions of the states of the unobserved regions compared to an empirical estimate that does not use the structural information, yet it is on the same level as the estimate that takes the structure into account. Furthermore, a comparison with the performed surgical resection and the surgery outcome indicates a link between the inferred excitable regions and the actual epileptogenic zone. The results emphasize the importance of the structural connectome in the large-scale spatio-temporal organization of epileptic seizures and introduce a novel way to integrate the patient-specific connectome and intracranial seizure recordings in a whole-brain computational model of seizure spread.

Does the regulation of local excitation-inhibition balance aid in recovery of functional connectivity? A computational account.

Computational modeling of the spontaneous dynamics over the whole brain provides critical insight into the spatiotemporal organization of brain dynamics at multiple resolutions and their alteration to changes in brain structure (e.g. in diseased states, aging, across individuals). Recent experimental evidence further suggests that the adverse effect of lesions is visible on spontaneous dynamics characterized by changes in resting state functional connectivity and its graph theoretical properties (e.g. modularity). These changes originate from altered neural dynamics in individual brain areas that are otherwise poised towards a homeostatic equilibrium to maintain a stable excitatory and inhibitory activity. In this work, we employ a homeostatic inhibitory mechanism, balancing excitation and inhibition in the local brain areas of the entire cortex under neurological impairments like lesions to understand global functional recovery (across brain networks and individuals). Previous computational and empirical studies have demonstrated that the resting state functional connectivity varies primarily due to the location and specific topological characteristics of the lesion. We show that local homeostatic balance provides a functional recovery by re-establishing excitation-inhibition balance in all areas that are affected by lesion. We systematically compare the extent of recovery in the primary hub areas (e.g. default mode network (DMN), medial temporal lobe, medial prefrontal cortex) as well as other sensory areas like primary motor area, supplementary motor area, fronto-parietal and temporo-parietal networks. Our findings suggest that stability and richness similar to the normal brain dynamics at rest are achievable by re-establishment of balance.