Fig. 1.

Scheme of the data analysis procedure. (A) 19-channel scalp EEG recordings containing seizures are considered. (B) Cortical source mapping is performed using eLORETA. (C) 15 ROIs are studied by taking the first principal component from all sources within the regions. (D) Example time series of the ROIs reconstructed from the signals displayed in (A). (E) Functional networks are inferred from the signals of the ROIs using the PLV. (F) A computational model of ictogenicity (the theta model) is employed to simulate dynamics on the networks. (G) Example times series generated using the theta model on the network (E). (H) The NI is computed by measuring the impact of removing nodes on the network’s ability to generate seizures in silico. (I) The ROI with the highest NI is identified (colored blue) and the prediction is compared with intracranial electrode implantation (black dots), performed surgery and postsurgical outcome (metadata not represented here). The comparison consists of observing whether the ROI with highest NI is in the same hemisphere where surgery was performed, and whether it is concordant with intracranial electrode placement. The aim is to observe whether this framework could have added value to the clinical decision-making process of defining where to implant intracranial electrodes to map the epileptogenic zone. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)