Classification of Distinct Vividness of Auditory Imagery in the Brain Based on Support Vector Machine

Abstract

Neuroimaging studies have demonstrated that neural response of auditory imagery is related to the vividness of the imagery, while whether the different vividness of auditory imagery in the human brain can be classified is unclear. In the present study, the classification of fMRI signal between high- and low-vivid groups was implemented based on support vector machine. Results showed that the classification accuracy was significant and several brain regions could reflect the vividness of auditory imagery, including left inferior parietal lobe, right anterior superior temporal gyrus and middle temporal gyrus. Besides, there was no significant differences between classifications with liner kernel and radial basis function, suggesting the results are stable and independent of kernel functions.