Abstracts

Rationale: Focal epilepsy in children is often refractory to medications and associated with significant developmental problems. Diagnosis and determination of epileptogenic zone is important for understanding the mechanism of epilepsy and developing effective treatment plans. EEG has been widely used in clinical to capture the electrophysiological neuronal activity and to identify the pathologically abnormal brain. Recent progress in source imaging has advanced the usage of EEG in clinical settings. However, the source imaging of seizures remains challenging due to the dynamic characteristics of seizures and the EEG contamination of motion artifacts. In this study, we investigate the feasibility of using a dynamic seizure imaging approach to help noninvasively identify the seizure sources. Methods: Nine pediatric patients with medically intractable partial epilepsy were studied. The pre-operative scalp EEG was recorded during a long-term video EEG monitoring. All the patients had pre-operative high-resolution MRI scans and the MRI images were used to build patient-specific boundary element head models. A dynamic seizure imaging approach was utilized to extract the spatiotemporal ictal features from independent component analysis and to image the seizure sources in the cortex. All patients had resective surgery and the post-operative MRI images were acquired from the patients. Seven of the patients underwent intracranial EEG (iEEG) monitoring before the surgery. The seizure source imaging results were evaluated by comparing them with the surgically resected region and the seizure onset zone (SOZ) of intracranial recording. Results: The seizure source imaging results in the nine patients are concordant with the surgical resection. The estimated source maxima in 7 out of the 9 patients are within the surgically resected regions and the results in the other 2 patients are in close vicinity to the resection boundary. Six months after the surgery, five patients have been seizure free and four patients have seizure reduction outcome. In the 7 patients with intracranial recording available, the seizure imaging results are concordant with the seizure onset zone marked from iEEG. The estimated source maxima in 5 patients are around 1 electrode distance to the SOZ and the sources in the remaining 2 patients are 2~3 electrode distances to the iEEG-SOZ. Conclusions: Noninvasive imaging of seizure activity shows promise for the epilepsy treatment. Our study suggests that the spatiotemporal ictal features can be extracted and used to image the dynamics of seizure activity. The imaged seizure sources of the patients are co-localized with the surgically resected regions and seizure onset zones determined by intracranial monitoring. These results show the feasibility of utilizing EEG seizure imaging to noninvasively localize seizure sources and suggest its potential application for the pre-surgical planning of epilepsy patients.