z-logo
Premium
Artifact cleaning of motor imagery EEG by statistical features extraction using wavelet families
Author(s) -
Nagabushanam Perattur,
George Selvaraj Thomas,
Dolly Devaraj Raveena Judie,
Radha Subramanyam
Publication year - 2020
Publication title -
international journal of circuit theory and applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.364
H-Index - 52
eISSN - 1097-007X
pISSN - 0098-9886
DOI - 10.1002/cta.2856
Subject(s) - wavelet , daubechies wavelet , computer science , artificial intelligence , pattern recognition (psychology) , electroencephalography , artifact (error) , discrete wavelet transform , sampling (signal processing) , wavelet transform , standard deviation , mathematics , speech recognition , computer vision , statistics , filter (signal processing) , psychology , psychiatry
Summary Electroencephalogram (EEG) and its sub‐bands represent electrical pattern of human brain. EEG signal contains transient components, spikes, and different types of artifacts due to eye blinking, movement of the person, anxiety, and so forth, during EEG capture. Wavelet transforms are powerful mathematical tool for sampling approximation to get clean EEG. It also helps in filtering, sampling, interpolation, noise reduction, signal approximation and signal enhancement, and feature extraction. In this paper, we have analyzed artifact cleaning via PSD graphs and statistical features extracted from motor imagery EEG‐like standard deviation variance. For this, we considered 19 channels EEG signal and applied orthogonal Daubechies wavelet, bi‐orthogonal rbio wavelet and Coifman wavelets to check the better performance of different wavelets. Coifman wavelet uses both scaling function and vanishing moments for sampling approximation and hence give smooth sampling compared to rbio and Daubechies wavelet transforms. Coif is a compactly supported wavelet system which also helps in smooth sampling approximations than other wavelets in the state of arts. The detailed coefficients and approximate coefficients can be further used for extracting features from EEG and classification purposes. Artifacts cleaning is thus observed better in coif wavelet analysis compared to other wavelets from the power distributions as power spectral density (PSD) graphs, standard deviation and variance obtained. Matlab R2013b is used for filtering and sampling EEG. Python 2.7 is used for statistical features extraction.

This content is not available in your region!

Continue researching here.

Having issues? You can contact us here