Continuous wavelet transform for the detection and classification of epileptiform activity in the EEG

Abstract

Outlines a novel approach to wavelet based detection of epileptiform activity in the EEG. A special complex-valued wavelet filter is used in a continuous wavelet transform (CWT). The response of wavelet coefficients (WCs) to epileptiform discharges (EDs) is measured with respect to artifact-free background activity (BG). A detector based on the WCs of a single scale could operate at 45% sensitivity without false alarms or at 99% sensitivity with 19 false alarms per minute on an artifact-free recording.