Detection and classification of fast ripples using wavelets

Fast ripples (FRs) are hypothesized to be a biomarker of epileptogenic processes. In this communication, we introduce a two-step procedure for automatically detecting and classifying FRs. In the first step, we detect all events of interest (EOIs) in the frequency band ranging from 250 Hz to 600 Hz. Then, based on wavelet transform, a local energy vs frequency analysis is performed to assign each EOIs to a specific class: FRs, interictal epileptic spikes (IESs), and artifact. The results obtained in the context of real depth-EEG signals (human and animal) show high performance in term of sensitivity and specificity.