The Medical Neurology Branch, NINDS, and the Computer Systems Laboratory, DCRT, have collaborated on a research project to noninvasively localize epileptic discharge sources within the human brain by using neuromagnetic recording in conjunction with conventional electroencephalogram (EEG) recording. Many patients with seizure disorders exhibit low-level cellular discharges between seizures, indicated by interictal spikes or sharp waves in their EEG and magnetoencephalogram (MEG) recordings. This project involved the development of computer techniques for automating and enhancing the procedure that is used by NINDS neurologists to determine the intracranial locations of the sources of epileptiform discharges in patients with epilepsy. In a previous fiscal year, CSL designed a computer system that detects epileptiform discharges from the EEG and MEG signals in real-time. This system has been implemented with a number of detection algorithms, including both those developed by CSL and some selected from the published literature, and a variety of options for configuring these algorithms allowing the medical staff to choose the optimal method for a given patient. The system provides a real-time display of the signals showing where an event is detected and allows the neurologists to manually save both epileptiform discharges and longer seizure activity. During the past year, CSL performed a comprehensive clinical evaluation of the real-time detection system with a variety of patient signals and compared its performance to that of a new commercial system that uses an algorithm not available on the CSL system. The CSL system reliably detected epileptic events, performed better than the commercial system, and was placed into clinical operation.
