Rationale:
High-density EEG (HD-EEG) source localization enhances presurgical evaluation by improving spatial precision in mapping interictal epileptiform discharges (IEDs). Sleep is known to increase IED yield and also to impact localization reliability. This study investigates clustering of spikes across different vigilance states and their concordance with seizure onset zones.
Methods:
Recordings from consecutive patients with focal epilepsy, who underwent between 20 and 24 hours of recording using 256-channel HD-EEG (EGI Geodesic® Sensor Net) with simultaneous video and physiological monitoring, over a 10-month period, were analysed. Spikes were marked manually, grouped into different populations (A–D), and categorized by sleep stage (Awake, N1, N2, N3, REM). Source localization was performed using sLORETA on individualized head models derived from patient-specific Magnetic Resonance Imaging (MRI) images. Clustering was assessed qualitatively based on consistency of anatomical localization across spike populations. Seizure onset zones, when available, were separately analyzed.
Results:
During the study period, 10 patients (4 females; mean age 42 士 11.07 years) underwent HD-EEG recording, mainly to strengthen the hypothesis of localization non-invasively. Analyzable data was available from 7 of these. A total of 400 spikes and 3 seizures were recorded. 34.5% of these spikes were recorded during wakefulness, while 65.5% occurred during sleep [14.25% in stage N1, 47.75% in stage N2, and 3.5% in stage N3 sleep]. Spike populations with ≥80% source threshold — highlighting regions with the highest likelihood of spike generation localizing to a single anatomical region — indicating high clustering—were observed in 100% of populations recorded during N1 sleep, 80% during N2 sleep, 67% during N3 sleep, and 43% during wakefulness.
Among individual patients, approximately 58% of spikes recorded during wakefulness were concordant with the proposed hypothesis for epileptogenic focus (determined through multidisciplinary team meetings collating data from clinical history, interictal/ictal EEG and video data along with MRI lesion location), compared with approximately 86% recorded during NREM sleep.
Conclusions:
These results from 256-channel high-density EEG source localization for patients with refractory focal epilepsy reveal sleep data to be generally more concordant with a proposed hypothesis for intracranial electrode implantation planning, especially among patients who are MRI negative.
Funding:
Lisa Calder, Tim Woodford, Helen Driver for technological support; Ann Norton and Janelle Bailey for administrative support, Sandra Weatherby and Melissa Meisner for nursing support.