Abstract
Individual differences in the size and shape of area V1 require that retinotopic mapping (RM) be performed on subjects before proceeding to an experiment. Despite variations in volumetric localization, V1 can be accurately identified using cortical surface anatomy (OP Hinds, 2008. Neuroimage 39:1585-1599). Using surface anatomy as a normalization guide, we asked whether the specific pattern of RM within V1 can be predicted for any given subject by anatomical features alone. We performed retinotopic mapping with BOLD fMRI (6 scans, 120 TRs, TR = 3 s, 3 mm voxels, at 3 Tesla) upon 10 subjects using both phase-encoded (PE) and multifocal (MF) methods (these data combined here). MPRAGE images (1mm isotropic voxels) for each subject were processed with FreeSurfer to identify and digitally inflate the cortical surface. The spherical cortical surface from each subject was then registered to an atlas and area V1 identified. The eccentricity and polar angle values were combined and then compared with data from each individual subject. Smooth, continuous retinotopic mappings within V1 were obtained such that all data points were resolved with standard deviations <2° of eccentricity and <30° of polar angle. Cross-validation of each individual subject to the template produced from the remaining 9 found a mean correlation of 0.48 ± 0.07 for polar angle and 0.28 ± 0.08 for eccentricity. The high agreement of the data between subjects as indicated by the small standard deviations demonstrates that reasonable estimates of V1 RM can be derived from anatomical features alone. Accurate prediction of the RM in V1, based purely on cortical surface anatomy, can greatly simplify studies of the visual cortex in healthy and patient populations. Future work will examine the reliability of retinotopic mapping from surface anatomy outside of area V1.
Supported by NIH Grants RO1 EY10016, PA Dept Health CURE 2010-06-01.