Evaluation of Different Wavelet Functions Applied in the Development of Digital Filters to Attenuate the Background Activity in EEG Signals
In this paper is presented a study where were evaluated five families of Wavelet function (Coiflets,Daubechies,Symlets,Biorthogonal and Reverse Biorthogonal) totalizing 65 evaluated functions and were used 500 epileptiform events for all the experiments.The purpose of this study was find out which Wavelet function is more appropriate to isolate specific frequencies that characterize the background activity of the EEG,generating a filtered signal only with the frequencies that characterize the epileptiform events.Several experiments were made,evaluating the original signals and the filtered signals and comparing its morphology through the correlations and Mean Squared Errors.Analyzing the obtained results from 65 Wavelet functions,three of them stood out: Rbio2.8,Db4 and Bior3.1 and both showed good attenuation of the background activity.However,each of them presented a particularity in relation to the morphology of the filtered signal.The function Rbio2.8 is more suitable to preserve the peak morphology of the epileptiform events.The function Db4 is more suitable to preserve whole epoch of signal that contains the epileptiform event.The function Bior3.1 presents good performance,due to the reduce number of coefficients,although it presents more distortion in the epileptiform events than the Rbio2.8 and Db4.
EEG Signal Processing Wavelet
G.R. Scolaro F.M. Azevedo C.F. Boos
Institute of Biomedical Engineering, Federal University of Santa Catarina, Florianopolis, Brazil
国际会议
World Congress on Medical Physics and Biomedical Engineering (2012年医学物理及生物医学工程国际会议(IFMBE))
北京
英文
340-343
2012-05-26（万方平台首次上网日期，不代表论文的发表时间）