Detection and localization of epileptic brain activity using an artificial neural network for dipole source analysis

We present a method for the detection of focal epileptic spikes in the EEG (electroencephalogram). The method is based on the dipole source localization technique and provides a source location estimate for each detected spike. An artificial neural network is used for performing dipole source localization, in order to be able to perform the detection method in real-time. The sensitivity and specificity of the method is studied on the basis of real patient EEG. It is observed that the method is able to operate with a sensitivity and specificity of 80% and 60%, respectively. We conclude that the method is suitable for real-time preprocessing of long-term EEG recording, in order to extract the relevant EEG epochs for subsequent source analysis.