Myelin densities in retinotopically defined dorsal visual areas of the macaque

Abstract

AbstractThe visuotopic organization of dorsal visual cortex rostral to area V2 in primates has been a longstanding source of controversy. Using sub-millimeter phase-encoded retinotopic fMRI mapping, we recently provided evidence for a surprisingly similar visuotopic organization in dorsal visual cortex of macaques compared to previously published maps in New world monkeys (Zhu and Vanduffel, Proc Natl Acad Sci USA 116:2306–2311, 2019). Although individual quadrant representations could be robustly delineated in that study, their grouping into hemifield representations remains a major challenge. Here, we combined in-vivo high-resolution myelin density mapping based on MR imaging (400 µm isotropic resolution) with fine-grained retinotopic fMRI to quantitatively compare myelin densities across retinotopically defined visual areas in macaques. Complementing previously documented differences in populational receptive-field (pRF) size and visual field signs, myelin densities of both quadrants of the dorsolateral posterior area (DLP) and area V3A are significantly different compared to dorsal and ventral area V3. Moreover, no differences in myelin density were observed between the two matching quadrants belonging to areas DLP, V3A, V1, V2 and V4, respectively. This was not the case, however, for the dorsal and ventral quadrants of area V3, which showed significant differences in MR-defined myelin densities, corroborating evidence of previous myelin staining studies. Interestingly, the pRF sizes and visual field signs of both quadrant representations in V3 are not different. Although myelin density correlates with curvature and anticorrelates with cortical thickness when measured across the entire cortex, exactly as in humans, the myelin density results in the visual areas cannot be explained by variability in cortical thickness and curvature between these areas. The present myelin density results largely support our previous model to group the two quadrants of DLP and V3A, rather than grouping DLP- with V3v into a single area VLP, or V3d with V3A+ into DM.