Improved Searchlight Method and the application in fMRI

• 1、College of Electronics and Information Engineering, Tongji University

Abstract：The brain with complex structure has been studied for many years. The studies about brain are not only focused on the anatomical structure, but also the function of brain nowadays, because of the development of the neuroimaging technique. Function Magnetic Resonance Imaging (fMRI) has advantages of being not traumatic to human body, high resolution of time and space and being easily conducted repeatedly. fMRI has been widely used in the field of cognitive psychology.Recent, machine learning has been introduced into fMRI data analysis. This kind of multivariate method computes not only the relation of the time series but also the space relation of different voxels and is more sensitive to the space relations.This paper introduces frequently-used multivariate methods especially "Searchlight" which can locate an area in the whole brain using multivariate methods. We improved Searchlight by introducing the algorithm of recursive feature elimination (RFE) into origin Searchlight. And prove the advantages of the improved method by several experiments.

Keywords： Pattern Recognition fMRI Multivariate Analysis