An extended convolution dynamic model of fMRI BOLD response

Abstract

Blood oxygenation level dependent (BOLD) contrast based functional magnetic resonance imaging (fMRI) can be used to detect brain neural activities and to reveal hemodynamic characteristics of various brain regions activated by a stimulus. In this paper, an extended convolution dynamic model of a BOLD signal is proposed as a new BOLD model for the interpretation of the fMRI signal. The model introduces a convolution between a cerebral blood flow kernal function and the perfusion function of neural response to a stimulus. Moreover, our model has extended previous models by adding a baseline and a shift of the baseline. In a visual evoked fMRI experiment, the parameters of the model are estimated with a nonlinear optimal algorithm. The results show that the extended convolution model fits the fMRI data well, and that the parameters are different in the left and right occipital region of the primary visual cortex. This finding indicates that the dynamic performances are different in various cerebral functional regions.