Fig. 6: (Patient 12) AEC-VIZ is discordant with MI-VIZ and variability is found across seizures.
From: Virtual intracranial EEG signals reconstructed from MEG with potential for epilepsy surgery
Dynamical network models identify different hotspots and boundaries between AEC-VIZ and MI-VIZ across two of the recorded seizures. MI-VIZ from seizure 1 gives an extensive boundary in the right frontal lobe with a hotspot that overlaps the resection margin, iEEG SOZ and early-ESL, which is the earliest solution that more accurately predicts the EZ than early-MSL (Supplementary Fig. 16). Boundaries and hotspots for the corresponding AEC-VIZ and MI-VIZ (seizure 2) hotspots are discordant with early-ESL, resection margin and iEEG SOZ. The patient is seizure-free over 24 months post-surgery and hence only MI-VIZ from seizure 1 successfully captures the EZ. Abbreviations: MEG magnetoencephalography, iEEG intracranial electroencephalography, ViEEG virtual intracranial electroencephalography, EZ epileptogenic zone, SOZ seizure onset zone, HDEEG high density electroencephalography, VIZ virtual ictogenic zone, MSL MEG source localisation, ESL HDEEG source localisation, AEC amplitude envelope correlation, MI mutual information, AEC-VIZ virtual ictogenic zone using amplitude envelope correlation, MI-VIZ virtual ictogenic zone using mutual information, NI node ictogenicity.
