Adding Transients to model BOLD fMRI Time Courses for Somatosensory-Motor Activations

Introduction: Within the General Linear Model (GLM) framework of analyzing fMRI data, an assumed neuronal input function is convolved with a hemodynamic response function (HRF) to model the fMRI time course. For stimulation of sensory-motor cortex, the simplest model of the neuronal input is a boxcar function with the same duration as the stimulus. Electrophysiological recordings, however, show that the associated neuronal activity commonly exhibits onset and offset transients [1,2,3]. Here, we investigate use of neuronal input functions with and without transients and variable HRF latency to explain the fMRI time courses in major centres of the somatosensory-motor system. The simple boxcar model, which implies linearity of the integrated BOLD response with respect to stimulus length, may work for some stimulus types and longer stimuli [4, 5]. However, non-linear deviations from this model have been observed and discussed in detail [5, 6] mostly for primary sensory cortices including the primary somatosensory cortex [7]. In this context, it is likely, that a model including transient components will fit fMRI data better.