Automatic analysis of interictal epileptic activity related to its morphological aspects

The objectives of this paper are to describe: (i) an algorithm, conventionally called TD (transient discriminator) which allows morphological discrimination among different EEG transients, and (ii) an automated methodology, based upon the proposed algorithm. The completely automated analysis performed by this method can indicate the temporal distribution of a 'family' of spikes, a family being defined by the morphology of a sample spike. Temporal relationships among the detected spikes of the family and other possible epileptic potentials arising from different cerebral zones are automatically defined by an averaging technique. The reliability of the proposed algorithm was investigated comparing the computer detection with the identification performed by 3 electroencephalographers.