EEG-FMRI integration using a partially constrained tensor factorization

Simultaneous EEG-fMRI recording provides great opportunity to study the relationship between synchronous neuronal activity in EEG and blood oxygenation level dependent (BOLD) in fMRI. In this paper a novel semi-blind technique using PARAFAC2 is proposed to investigate the correlation between post-movement beta rebound (PMBR) in beta band and BOLD. In this method, the instantaneous power of EEG in beta band representing the PMBR is calculated and used as a constraint for PARAFAC2 to detect the fMRI voxels activated during PMBR. The results confirm that the proposed method effectively detects the area in the brain which is responsible for beta rebound. The results are compared with those of general linear model (GLM) that completely relies on the predefined fMRI time-course.