Source Localization of Simulated Electroencephalogram of Virtual Epileptic Patient to Investigate Clinically Feasible Montages

Electroencephalogram (EEG) source localization is used to estimate regions of ictal onset in epilepsy patients with temporal lobe epilepsy. Localization of EEG data struggles to achieve high spatial resolution, especially in deep brain regions, and is difficult to validate. In this paper we generate simulated EEG data using a spatiotemporally realistic generative brain network model (BNM) based on patient structural and functional data, created with The Virtual Brain (TVB) platform, to qualitatively assess head model approaches, distributed source inverse methods and clinically feasible electrode montages. We find that sLORETA is highly sensitive to head model errors, where dSPM is robust and wMNE displays some sensitivity. Additionally, increased electrode density over regions of interest provides a clinically feasible means to improve localization accuracy of a sparse montage. Finally, TVB platform can be utilized to model patient anatomy and physiology where resultant simulated EEG can be source localized for personalized neurological care.