How Much Does the Reduced EEG Montage Matter for Seizure Detection?: A Large-Cohort Simulation Study

Area of Science:

• Epilepsy monitoring and diagnostics
• Biomedical engineering
• Signal processing in neuroscience

Background:

• Implantable sub-scalp electroencephalogram (EEG) systems offer promising long-term seizure monitoring for epilepsy patients.
• The impact of different montage configurations on seizure detection performance in these systems is not well understood.

Purpose of the Study:

• To quantify differences in automated seizure detection performance between full and reduced EEG montages.
• To investigate how these differences vary based on specific epilepsy characteristics.

Main Methods:

• Retrospective cross-sectional study analyzing EEG data from 466 patient admissions.
• Computational simulation of published sub-scalp device montages using standard 10-20 EEG channels.
• Evaluation of three seizure detection algorithms (SVM, SPaRCNet, NDD) using event-based F1 scores.

Main Results:

• Seizure detection performance showed only modest decreases (absolute F1 change -0.09 to 0.014) with reduced montages compared to full montages.
• Patient factors (29.2%) and algorithm choice (10.3%) accounted for the majority of performance variance, with minimal montage effects (0.4%).
• Performance on reduced montages moderately to highly correlated with full-montage performance (ρ=0.29-0.73).

Conclusions:

• Automated seizure detection is feasible with reduced-montage sub-scalp devices, achieving comparable accuracy.
• Detector and patient-level factors are more critical than montage configuration for seizure detection performance.
• Optimizing long-term performance may require patient-specific seizure detection algorithms and careful patient selection.