Fragility in epileptic networks: The epileptogenic zone

Abstract

Epilepsy is a network phenomenon characterized by seizures that cause hypersynchronous activity of populations of neurons at the cortical level. In our previous work, we defined a seizure event as an outcome of a stable networked system switching to an unstable system. More specifically, we argued that the epileptic network is on the brink of instability, and that small perturbations to synaptic connections between populations of neurons (network nodes) render the stable network unstable, causing a seizure. Fragility of each network node was defined and computed for linear network models. Here, we applied the theory to the problem of identifying where seizures start from, defined as the epileptogenic zone (EZ), given intracranial electrocorticographic (ECoG) recordings. We hypothesize that the most fragile nodes of an epileptic network correspond to the region of the brain that start seizures. We test our hypothesis by (i) estimating linear models from ECoG recordings obtained from two epilepsy patients, (ii) computing the fragility of each electrode contact (network node), and (iii) by comparing the most fragile nodes as derived from our model for each patient to clinically annotated EZ. We found that the most fragile nodes that destabilize the epileptic cortical network in each patient have a high degree of concordance with the clinical annotations and can thus be used to help identify the EZ which currently is a difficult, costly and time consuming process.