A new approach to spike detection using a combination of inverse and matched filter techniques

Abstract

For the detection of spikes and other types of transient signal, a combination of inverse and matched filter techniques can be used. The inverse filter improves the signal to noise (spike to EEG) ratio; the matched filter makes it possible to classify different types of spikes. When compared, the combined inverse matched filter technique gave a higher detection rate than the inverse filter technique alone, for the same error probability. Both simulated and patient data were used for testing the methods.