Abstract
During development, retinal hypo-pigmentation in albinism leads to aberrant decussation of temporal retinal afferents at the optic chiasm thereby producing partially superimposed representations of opposite hemi-fields in visual cortex. Previous studies of neuronal mis-wiring in achiasmic patients found that population receptive fields (pRFs) in visual cortex were often split into two separate regions typically positioned at mirror image locations across the vertical meridian (Hoffmann et al., 2012). However, such dual-pRFs have not been reported or quantitatively modeled in albinism. We used conventional ring and wedge stimuli with fMRI at 3T to map cortical retinotopy and model individual voxel pRFs throughout the visual hierarchy in 5 albinism subjects. Our hypothesis was that voxel responses within hemi-field overlap regions would be fit best by dual rather than single gaussian models. This is consistent with the qualitative observation that single voxel responses in overlap zones often display twice the number of expected response peaks to rotating wedge stimuli. To test our hypothesis, we first used monocular, left and right hemi-field ring stimuli to identify cortical zones with overlapping representations of opposite hemifields. Consistent with previous reports, all albinism subjects exhibited aberrant regions of hemi-field overlap. Voxel responses throughout visual cortex were fit with both single and dual gaussian pRF models. To compare the model fits, we used the residual sum squared error and the number of model parameters to compute Akaike's information criterion. Of nearly 11,000 voxels successfully modeled, over 1000 were fit best by the dual pRF model, and the majority of these fell within overlap zones. In contrast, of ~7,000 voxels in non-overlap zones the majority were fit better with a single gaussian model. These results provide quantitative evidence that gross aberrations of cortical retinotopic organization in albinism are further reflected in aberrant pRF structure at the level of single voxels.
Meeting abstract presented at VSS 2018