Effects of inverse methods and spike phases on interictal high-density EEG source reconstruction

To determine the effect of inverse methods and timepoints of interictal epileptic discharges (IEDs) used for high-density electric source imaging (hd-ESI) in pharmacoresistant focal epilepsies.

We retrospectively evaluated the hd-ESI and [18F]fluorodeoxyglucose positron emission tomography (¹⁸FDG-PET) of 21 operated patients with pharmacoresistant focal epilepsy (Engel I). Volumetric hd-ESI was performed with three different inverse methods such as the inverse solution linearly constrained minimum variance (LCMV, a beamformer method), standardized low resolution electromagnetic tomography (sLORETA) and weighted minimum-norm estimation (wMNE) and at different IED phases. Hd-ESI accuracy was determined by volumetric overlap and distance between hd-ESI source maximum, as well as ¹⁸FDG-PET hypometabolic region relative to the resection zone (RZ).

In our cohort, the shortest distances and greatest volumetric overlaps to the RZ were found in the half-rise and peak-phase for all inverse methods. The distance to the RZ was not different between the centroid of the clinical hypothesis-based cluster and the source maximum in peak-phase. However, the distance of the hypothesis-based cluster was significantly shorter compared to the cluster selected by the smallest p-value.

Hd-ESI provides the greatest accuracy in determining the RZ at the IED half-rise and peak-phase for all applied inverse methods, whereby sLORETA and LCMV were equally accurate.

Our results offer guidance in selecting inverse methods and IED phases for hd-ESI, compare the performance of hd-ESI and ¹⁸FDG-PET and encourage future studies in investigating the relationship between interictal ESI and ¹⁸FDG-PET hypometabolism.