The Use Of Multi-channel Information In The Detection Of Epileptiform Activity In The EEG

A PC-based system has been developed to automatically detect epileptiform events in the inter-ictal EEG. The system consists of three stages : data collection, data reduction and confirmation and classification of epileptiform events. A basic multi-channel capability has been introduced into the data reduction stage (in the form of a two threshold system) to increase the proportion of epileptiform transients detected. The classification of epileptiform events as definite or probable overcomes, to an extent, the problem of maintaining high detection rates while eliminating false detections. The system has been evaluated on the EEGs of 8 patients with 54-70% of epileptiform events being detected as definite (i.e no false detections), and 60-100% as either definite or probable events at the expense of introducing up to 15 false detections per hour. Current work aimed at taking temporal, as well as spatial, information into account should realize still higher detection rates.