Computing the Time-dependent Krankheit-operator in Epilepsy from ECoG: a Case Study

Description and prediction of epileptic seizures present numerous challenges that can be addressed with physics and computer science. Here, we investigate epilepsy by developing a specific form of the Krankheit-operator (K-operator), a physics-based approach modeling disease-driven damage on brain pathways. The K-operator acts on the different layers of the brain, from neurons, to neural agglomerates, to lobes. Its first experimental applications to functional magnetic resonance images dealt with interactions between regions of interest of the brain. Here, we consider the action of K between different brain areas described by channels in electrocorticography (ECoG) for the first time. In particular, we focus on temporal lobe epilepsy, applying the methodology to a case study, i.e., the data acquired on a person monitored via pre-surgery ECoG. The information before, immediately before, during, and after an epileptic seizure is encoded in matrices, and investigated with the tools of operatorial algebra adopted in physics, shaping a form of the K-operator for our case study. We discuss our preliminary results and sketch further lines of development.