Controlling seizure propagation in large-scale brain networks

Abstract

Information transmission in the human brain is a fundamentally dynamic network process. In partial epilepsy, this process is perturbed and highly synchronous seizures originate in a local network, the so-called epileptogenic zone (EZ), before recruiting other close or distant brain regions. We studied patient-specific brain network models of 15 drug-resistant epilepsy patients with implanted stereotactic electroencephalography (SEEG) electrodes. Each personalized brain model was derived from structural data of magnetic resonance imaging (MRI) and diffusion tensor weighted imaging (DTI), comprising 88 nodes equipped with region specific neural mass models capable of demonstrating a range of epileptiform discharges. Each patient’s virtual brain was further personalized through the integration of the clinically hypothesized EZ. Subsequent simulations and connectivity modulations were performed and uncovered a finite repertoire of seizure propagation patterns. Across patients, we found that (i) patient-specific network connectivity is predictive for the subsequent seizure propagation pattern; (ii) seizure propagation is characterized by a systematic sequence of brain states; (iii) propagation can be controlled by an optimal intervention on the connectivity matrix; (iv) the degree of invasiveness can be significantly reduced via the proposed seizure control as compared to traditional resective surgery. To stop seizures, neurosurgeons typically resect the EZ completely. We showed that stability analysis of the network dynamics, employing structural and dynamical information, estimates reliably the spatiotemporal properties of seizure propagation. This suggests novel less invasive paradigms of surgical interventions to treat and manage partial epilepsy.

Temporal Lobe Epilepsy Surgical Outcomes Can Be Inferred Based on Structural Connectome Hubs: A Machine Learning Study

“…There is a robust overlap between abnormally strong connections and increased "centrality" of ictal‐onset zones to sites of noncontiguous spread in the structural connectome along with increased amplitude of corticocortical evoked potentials to those same distant sites (effective connectome) 16 . Seizure freedom is seen in association with statistical models that virtually resect hyperconnected 17 and hypersynchronized 18 network hubs. Moreover, postoperatively, node‐based measures of centrality rather than global network measures are associated with postoperative outcomes 19 .…”

Neurostimulation stabilizes spiking neural networks by disrupting seizure-like oscillatory transitions

“…In this vein we sought to understand the effects of neuromodulation on seizure-like transitions from a computational and mathematical perspective, allowing a principled understanding of dynamical interactions between modeled neurons and stimulus parameters. There are numerous existing computational models of ictogenesis 9 , 22 – 25 and seizure propagation 26 – 28 . Irrespective of the different neurophysiological mechanisms involved in ictogenesis 29 , these in silico models generally represent seizure onset via spontaneous transitions between irregular spiking dynamics and hyperactive oscillatory states.…”

Patient-specific network connectivity combined with a next generation neural mass model to test clinical hypothesis of seizure propagation

“…This form of virtual brain modeling Jirsa et al (2002Jirsa et al ( , 2010 exploits the explanatory power of network connectivity imposed as a constraint upon network dynamics and it has been mostly applied, so far, to uncover basic principles of brain network functioning by making use of generic or averaged connectomes. Recent studies have pointed out the influence of individual structural variations of the connectome upon the large-scale brain network dynamics of the models Proix et al (2017Proix et al ( , 2018; Olmi et al (2019), by systematically testing the virtual brain approach along the example of epilepsy. The employment of patient-specific virtual brain models derived from diffusion MRI may have a substantial impact for personalized medicine, allowing for an increase in predictivity with regard to the pathophysiology of brain disorders, and their associated abnormal brain imaging patterns.…”

“…In order to exploit the predictive power of personalized brain network models we have implemented, on patient-specific connectomes, a next generation neural mass model that, differently from the previous applied heuristic mean-field models Proix et al (2017Proix et al ( , 2018; Olmi et al (2019), is exactly derived from an infinite size network of quadratic integrate-and-fire neurons Montbrió et al (2015), that represent the normal form of Hodgkin's class I excitable membranes Ermentrout and Kopell (1986). This next generation neural mass model is able to describe the variation of synchrony within a neuronal population, which is believed to underlie the decrease or increase of power seen in given EEG frequency bands while allowing for a more direct comparison with the results of electrophysiological experiments like local field potential, EEG and event-related potentials (ERPs), thanks to its capacity of capturing the macroscopic evolution of the mean membrane potential.…”