Abstract
The human brain contains many different regions defined by sensory neural responses and anatomical landmarks. FMRI allows for non-invasive investigation of how various sensory inputs correspond to their cortical representations. Accurate and reliable identification of these cortical regions in individual subjects is essential for group-level analyses, but individual delineation of regions of interest (ROIs) is not always feasible in studies with large datasets or missing data. In this work, we examined the retinotopic fMRI responses within visual cortex to assess the general utility of a publicly available probabilistic atlas (Wang et al., 2015) for independently identifying three important topographic areas in the visual system: V1, V2, and V3. We modified a standard population receptive field (pRF) mapping protocol by systematically modulating the stimulus flicker frequency and the types of objects that composed the bars of visual stimuli sweeping across the visual field. We also included motor cues and auditory stimuli in the pRF protocol. Twenty-five neurotypical adults completed two five-minute pRF scans at 7 Tesla (1.6 mm isotropic resolution). Robust motor, auditory, and retinotopic mapping were observed in the expected cortical areas. In our sample, we found reliable agreement between independent V1 identification based on structural MRI data and the expected fMRI response. However, we found significant discrepancies at the individual subject level between the independent atlas definition of the V2 and V3 boundaries and the fMRI response. These results emphasize the continued need for visual inspection of ROIs in visual cortex and validate a protocol that allows efficient and automated mapping of visual, auditory, and motor representations.