EEG based Motor Imagery BCI using Four Class Iterative Filtering & Four Class Filter Bank Common Spatial Pattern

Motor Imagery Brain Computer Interface is the cognitive process and most momentary brain activity which operate various limb movements depends on EEG brain signals by just imagined or without tensing the muscles but MI BCI suffers from poor performance compared to P300 and SSVEP BCI so in this paper proposes an integrated Four Class-Iterative Filtering(FC-IF) and Four Class-Filter Bank Common Spatial Pattern (FC-FBCSP) uses the more popular technique CSP-PW(Piece Wise) and CSP-OVR(One Over The Rest) which improves the classification accuracy of 95.29% compared to all other previous CSP techniques. In this research paper the research is substantiated using freely accessible EEG data set BCI Competition IV-2(a) in which firstly the EEG preprocessing is done to remove the artifacts and non stationary noise and EOG artifacts removed by using Four Class-Iterative Filtering(FC-IF) after improving signal preprocessing applied FC-FBCSP to extract multiple frequency bands of the EEG signal and applied the CSP algorithm to each frequency band to enhance the spatial information of the EEG signals in multiple frequency bands. The significant features optimally selected by using proposed technique, which improves the spatio - temporal features in order to improve performance measurements of MI BCI which aim to improve the classification accuracy. Iterative filtering in which each iteration uses the outcomes of the prior iteration as input involves repeatedly recurring the decomposition and continued with FC-FBCSP filtering steps & with a variation of the CSP method. As a result, classification performance enhanced and the differences between the two classes of signals are identified. The popular Support Vector Machine(SVM) or Naïve Bayes (NB) classifier is used to classify the divided classes are class1 (Left hand), class2 (Right hand), class3(Foot), class4( Tongue). FC-IF & FC-FBCSP is a suitable choice for MI BCI application because of its outstanding classification results and improved accuracy.