Recognition of Epileptic EEG Using Probabilistic Neural Network

Epilepsy is one of the most common neurological disorders that greatly impair patients' daily lives. A classifier for automated epileptic EEG detection and patient monitoring can be very important for epilepsy diagnosis and patients' quality of life, especially for rural areas and developing countries where medical resources are limited. This paper describes our development of an accurate and fast EEG classifier that can differentiate the EEG data of healthy people from that of epileptic patients, and also detect patients' status (i.e., interictal vs. ictal). We deployed Probabilistic Neural Network (PNN) and fed it with 38 features extracted from the EEG data. The resulting PNN EEG classifier achieves an impressive accuracy greater than 96 as indicated by cross-validation. This prototype classifier is therefore suitable for automated epilepsy detection/diagnosis and seizure monitoring. It may even facilitate seizure prediction.