Abstract
Intro: PRF models fit to fMRI data are a component of many visual experiments. It is important to know how reliable pRF parameters are across experimental conditions. We compared retinotopic maps and pRF properties across two datasets obtained with different fMRI protocols and stimulus designs: a newly acquired dataset from New York University (NYU) (n=44) and a public dataset from the Human Connectome Project (HCP) (n=181). The datasets differ in many ways, including field strength (3T vs 7T) and voxel size (1.6 vs 2 mm). Method: PRF parameters (polar angle, eccentricity, and pRF size) were estimated using vistasoft software (https://github.com/vistalab/vistasoft) on group-averaged time-series (registered to the fsaverage surface) and on each individual subject’s time-series. Result: Within V1, V2, V3, and hV4, the vertex-wise polar angle estimates of the group-averaged data were similar between the two datasets (r=0.97, V1-hV4). The eccentricity estimates were also highly correlated (r=0.94) but with a small bias toward higher eccentricities in the NYU dataset. On average, pRF size estimates were correlated (r=0.81) and showed a greater systematic difference between datasets: They were larger (by 0.5 to 1º) in the NYU dataset across comparable eccentricities (0.5-6°) and visual areas. Finally, we examined a large-scale property of the maps in individual subjects, polar angle asymmetries in V1 cortical magnification. These were recently described in the HCP dataset (Benson et al., 2020): Here we show that the NYU dataset replicates the finding that more V1 cortical surface area represents the horizontal than vertical visual field meridian, and the lower than upper vertical meridian. Conclusion: The retinotopic maps (polar angle, eccentricity, cortical magnification) are consistent between the datasets, indicating a high degree of reproducibility despite numerous experimental differences. PRF sizes are roughly 1º larger in the NYU dataset, a difference that may be due to larger voxel sizes.