Coupling of simultaneously acquired electrophysiological and haemodynamic responses during visual stimulation

Abstract

We investigate the relationship between the temporal variation in the magnitude of occipital visual evoked potentials (VEPs) and of haemodynamic measures of brain activity obtained using both blood oxygenation level dependent (BOLD) and perfusion sensitive (ASL) functional magnetic resonance imaging (fMRI). Volunteers underwent a continuous BOLD fMRI scan and/or a continuous perfusion-sensitive (gradient and spin echo readout) ASL scan, during which 30 second blocks of contrast reversing visual stimuli (at 4 Hz) were interleaved with 30 second blocks of rest (visual fixation). Electroencephalography (EEG) and fMRI were simultaneously recorded and following EEG artefact cleaning, VEPs were averaged across the whole stimulation block (120 reversals, VEP(120)) and at a finer timescale (15 reversals, VEP(15)). Both BOLD and ASL time-series were linearly modelled to establish: (1) the mean response to visual stimulation, (2) transient responses at the start and end of each stimulation block, (3) the linear decrease between blocks, (4) the nonlinear between-block variation (covariation with VEP(120)), (5) the linear decrease within block and (6) the nonlinear variation within block (covariation with VEP(15)). VEPs demonstrated a significant linear time-dependent reduction in amplitude, both within and between blocks of stimulation. Consistent with the VEPs finding, both BOLD and perfusion measures showed significant linear time-dependent reductions in response amplitude between blocks. In addition, there were significant linear time-dependent within-block reductions in BOLD response as well as between-block variations positively correlating with VEP(120) (medial occipital and frontal) and within-block variations positively correlating with VEP(15) (occipital and thalamus). Both electrophysiological and haemodynamic (BOLD and ASL) measures of visual activity showed steady habituation through the experiment. Beyond this, the VEP measures were predictive of shorter timescale (3-30 second) localised variations in BOLD response engaging both occipital cortex and other regions such as anterior cingulate and parietal regions, implicating attentional processes in the modulation of the VEP signal.