Brain-Computer Interface for Patient to Command Left-Right Using Imagination by Wavelet Packet Filter

Abstract

This paper presents Hand Motor Imagery (HMI) using Electroencephalogram (EEG). The key idea is to useWavelet Packet Transform (WPT) to filter irrelevant signal; Comparing to using static filters. Band selection is basedon an objective function which is consisted of wavelet coefficients. Local Discriminant Bases (LDB) algorithm isadopted to select the most appropriate bands. Feature is variance of Common Spatial Pattern (CSP). Bayes DecisionTheory (BDT) is used as a signal classifier. The proposed method gains higher accuracy than many existing methods.