The study of data analysis methods based on FMRI brain-computer interface

Abstract

In this paper, we explore whether the functional magnetic resonance imaging(fMRI) signals of advanced brain areas except the motor area, prefrontal brain regions can be used in the brain-computer interface and how to effectively improve the accuracy of thinking classification by extracting blood oxygen level dependent(BOLD) signal features. We focus on whether the fMRI brain signal of posterior parietal cortex(PPC) can be used in brain computer interface. By using peak value and cumulative changes to select features, and using support vector machines(SVM) to classify data, we've drawn the conclusion that PPC brain regions can be well applied in brain computer interface and the classified accuracy using peak value is higher than the classified accuracy using cumulative changes.