Author:

Abstract:

Cytochrome oxidase stainings revealed three types of stripes in macaque and human area V2. The widths of these stripes are roughly 1 - 1.5 mm in macaque and 2.5 - 3.5 mm in human. Stripes of the same kind are seperated from each other by ~4 mm in macaque (Roe and Ts'o 1997) and ~7 mm in human (Nasr et al. 2016). Different functional properties are attributed to thin and thick dark stripes, for example based on separated color, direction, and disparity-selective cells (Chen et al. 2008, Tootell et al. 2004). Nevertheless, the stripes remain difficult to segregate using conventional fMRI due to their size. Recently, an fMRI study (Nasr et al. 2016) using high field (7T) and 1 mm isotropic voxels exquisitely revealed this stripe-like functional organization in human area V2. Here, we visualized the stripes in awake macaques using sub-mm resolution and conventional field fMRI (3T). We used the same isoluminant color and achromatic radial grating stimuli as in our previous double-label deoxyglucose study (Tootell et al. 2004) to detect the color-biased thin stripes. A size-matched binocular disparity defined radial sine-wave grating and its monocular counterpart was used to activate the disparity-biased stripes. Furthermore, myelination density mapping using T1- and T2-weighted MRI with 0.4 mm isotropic voxel size was acquired in an attempt to visualize the pale stripes which are more myelinated compared to thin and thick stripes (Horton and Hocking 1997). During the functional scans, monkeys fixated at the center of the stimulus. Contrastenhanced fMRI data with 0.6 mm isotropic voxels (EPI, TR 3 s) was acquired using implanted phased-array receive coils developed in our lab (Janssens et al. 2012). To minimize image artifacts caused by monkey motion, several image processing algorithms were implemented, including an optimized GRAPPA algorithm for image reconstruction (Hoge et al. 2009), a B-spline grid based nonlinear registration algorithm for EPI to EPI distortion correction and a fieldmap-based distortion correction algorithm for EPI to structue registration. Finally, a GLM denoising algorithm (Kay et al. 2013) was used to reduce physiological and other nuisance noise in the temporal domain. We found stripe-like color activations interdigitated with disparity activations in retinotopically-defined area V2, which are reproducible across monkeys and across scan sessions of the same monkeys. Furthermore, we observed higher myelinated stripes largely separated from color- and disparity-selective stripes. These results show that high-resolution 3T fMRI can be reliably used to study in-vivo submillimeter-scale functional organizations of the primate brain.