Online dictionary learning based on subject connectivity in fMRI data

Inhalt:
Functional magnetic resonance imaging (fMRI) based on the hemodynamic response has been widely used to investigate the functional connectivity relationships between the brain regions, which could uncover the neural correlates of cognitive processes. The previous studies mainly performed functional connectivity analysis from single subject first and then perform group analysis to infer group functional connectivity pattern from a group of subjects. However, due to the considerable variability among subjects, functional connectivity analysis cannot tell us whether the variation of subjects in one brain region is coupled with that in the other region. In order to reveal the correlation between subjects’ variations, the subject connectivity that is the correlation between the subjects’ variations of two different regions was defined in this study. We proposed an online dictionary learning framework based on subject connectivity (ODL-SC) in fMRI data analysis. The ODL-SC method was applied to the motor imagery training fMRI data. The experimental results revealed that the motor imagery training increased the subject connectivity between the motor areas and the other brain regions.