This paper proposes a digital filter based on wavelet transform and performs an investigation to find an appropriate wavelet function for filtering epileptiform events in EEG signals using the wavelet multiresolution analysis. We investigated five known families of wavelet functions (Coiflets, Daubechies, Symlets, Biorthogonal and Reverse Biorthogonal), totaling 65 functions. In the experiments were used 600 epileptiform events, covering frequencies between 5-25 Hz (40-200ms). The experiments consisted in comparing the morphology of the original signals of epileptiform events with the filtered epileptiform events, estimating the correlation coefficient and root mean square errors between them. Analyzing the results obtained from the experiments four wavelet functions stood out: RBio2.8, Db4, Coif4 and Bior3.1. All of them proved to be able to perform good attenuation in the EEG background activity. The functions Rbio2.8 and Coif4 are appropriate to preserve the peaks morphology of the epileptiform events. The Db4 function is most suitable for preserving the morphology of the whole epoch that contains the epileptiform event. The function Bior3.1 due to the low number of coefficients showed less processing time in relation to the other ones. However, the morphology of the filtered epileptiform events presents more distortion in relation to the functions Rbio2.8, Coif4 and Db4.
KEY WORDSEEG, epileptiform events, wavelet transform, wavelet filter.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.