HOME   |   OBJECTIVES   |   RESULTS  |   TEAM

    Executive Summary

The outcome of the surgical treatment for epilepsy depends on the accurate localization of the epileptogenic tissue. A significant number of patients require long-term intracranial recordings during the presurgical evaluation phase.
Our main objective is to provide an algorithm for identifying the epileptogenic tissue based on altered directed functional connectivity (effective connectivity) patterns and state-of-the-art machine learning techniques. Altered connectivity associated with the epileptic tissue has been recently reported, however no prior study systematically analyzes the whole-brain effective connectivity patterns of healthy and pathological brain structures. As effective connectivity is defined as "the influence one neural system exerts over another", we will use intracranial electrical stimulation to evoke iEEG potentials for assessing the effective connectivity in a large patient cohort (n?75). We will be using a proven methodology that we have recently introduced (Donos et al., Neuroimage, 2016) and will benefit from ongoing collaboration on the topic with centers from Neuroscience Institute - Grenoble and King's College - London.
Secondary objectives include:
1. A differential analysis on the changes in effective connectivity patterns induced by epilepsy. For each brain structure we will create two subsets of patients, so that the structure will be considered epileptogenic in the first subset, and healthy in the second. Using a modulation index that quantifies the number and the strength of the connections, we will compare the effective connectivity patterns of each brain structure, in both conditions. These quantitative measures will be used as features for the random forest classifier that will help us accomplish our main objective.
2. A physiological effective connectivity map, presented as a whole brain atlas, as well as individualized effective connectivity maps of both healthy and pathological structures.