Interictal electric source imaging with open‐source software and routine low‐density electroencephalography: A systematic comparison of four inverse methods in drug‐resistant epilepsy

Abstract

Abstract Objective This study was undertaken to assess the diagnostic accuracy of interictal electrical source imaging (ESI) performed using open‐source software and routine low‐density electroencephalography (LD‐EEG), comparing four inverse methods at two spike time points. Methods We retrospectively included patients with drug‐resistant epilepsy (DRE) who underwent epilepsy surgery, had at least 1 year of postoperative follow‐up, and underwent routine LD‐EEG. Interictal ESI was performed in Brainstorm using four inverse methods—standardized low‐resolution electromagnetic tomography (sLORETA), linearly constrained minimum variance (LCMV) beamformer, minimum norm (MN), and Equivalent Current Dipole (ECD)—at two time points: the half‐rising phase and the peak amplitude of averaged spikes. Concordance with the postsurgical cavity and Engel class I outcome were used as the reference standard for evaluating diagnostic performance. Intermethod concordance was assessed at the diagnostic, lobar, and sublobar levels. Results Seventy‐nine patients were included (Engel I, 72.2%). ESI at the half‐rising phase was significantly more accurate than at peak amplitude for sLORETA ( p = .012) and ECD ( p < .001). sLORETA at the half‐rising phase achieved the highest diagnostic performance (accuracy 81.0%, sensitivity 82.5%, specificity 77.3%), significantly outperforming MN ( p = .003). Diagnostic accuracy of sLORETA at the half‐rising phase was maintained across subgroups, including patients with extratemporal epilepsy (84.6%) and nonlesional magnetic resonance imaging (80.8%) and patients requiring stereo‐EEG (80.0%). Diagnostic agreement among the four methods was substantial, with sLORETA, LCMV beamformer, and ECD forming a high‐concordance cluster (Cohen κ = .842–.857). Agreement was substantial at the lobar level ( κ = .755) and moderate at the sublobar level ( κ = .465). Significance Interictal ESI performed with open‐source software and routine LD‐EEG achieves high diagnostic accuracy for localizing the presumed epileptogenic zone in DRE, with robust results across traditionally challenging populations. These findings support the broader clinical adoption of ESI as a valuable tool in presurgical evaluation.