Substitution of spatial filters from relaxation to motor imagery for EEG based brain computer interface

An offline analysis for electroencephalogram (EEG) based on brain computer interface (BCI) was implemented. Independent component analysis (ICA) was used for separating Mu rhythm in both hemispheres and for generating spatial filters derived from relaxation state and from motor imagery state. The main purpose was to study the substitution of spatial filters from relaxation state to motor imagery one. A motor imagery dataset with 9 subjects was used. In order to extract features from brain signals, power spectral density was calculated for the independent components chosen with equivalent dipole. The classification was evaluated using three classifiers: linear discriminant analysis (LDA), support vector machine (SVM) and k nearest neighbor (kNN). Paired t-test demonstrated that substituted spatial filters and spatial filters from motor imagery were not statistically different.