Abstracts

Rationale: Long-term EEG monitoring is a major component of the pre-surgical evaluation of medically intractable epilepsy. However, this procedure suffers from low spatial resolution, and alone is often inadequate for the precise delineation of the epileptogenic brain. We report a novel spatiotemporal source imaging approach for localizing and tracking seizure activity using long-term, 76-electrode scalp EEG monitoring (dense-array EEG) . Direct imaging of seizure sources and their dynamics may improve surgical management of intractable epilepsy, and make high density EEG imaging a pre-surgical evaluation tool in the epilepsy treatment. Methods: Five adult patients underwent continuous video EEG monitoring using 76 channels. They all had high-resolution MRI prior to the surgery. Four patients underwent resective surgery and all became seizure free. Post-operative MRIs were collected for these four patients. Three patients underwent intracranial EEG monitoring, including the one without resective surgery. In the source imaging analysis, ictal EEG segment was decomposed using independent component analysis (ICA). The scalp map of each independent component (IC) was used to estimate the 3D brain source distribution by solving an EEG inverse problem. The time course of each IC was used to compute a spectrogram, and components showing time-frequency evolution of ictal rhythmic discharges were selected. As an inverse process of ICA, we re-combined the IC source distributions and IC time courses in 3D source space, which resulted in a spatiotemporal estimation of the brain ictal source activity. Results: A total of 11 seizures were analyzed. The seizure onset zone (SOZ) was localized for each patient as the source distribution at the time of seizure onset. In each of the four patients who underwent resective surgery, significant overlap was observed between the estimated SOZ and the resected brain volume, and furthermore the maximal estimated source point was localized within the resected region. In the patient not undergoing resective surgery, the estimated SOZ was concordant with epileptogenic brain determined by intracranial EEG and SPECT. In the three patients with intracranial electrodes, the seizure onset and propagation pattern estimated by source imaging was concordant with the intracranial recording. Conclusions: The results obtained in this study have potential implications for the practice of pre-surgical evaluation of epilepsy. The feasibility of obtaining ictal recordings from a 76-channel EEG montage was successfully demonstrated in a cohort of 5 patients. The clinical data was recorded over 6 4 days, and the 76-electrode montage successfully captured each seizure in this cohort of patients. With dense-array EEG the proposed seizure imaging technique makes possible direct imaging of the generators of seizures and their propagation. With the capability of recording and imaging ictal activity, the proposed high-density recording and electrographic seizure imaging technique provides a non-invasive imaging tool to assist pre-surgical planning in the treatment of medically intractable epilepsy.