Multi-resolution classification analysis of ocular dominance columns obtained at 7 Tesla from human V1: mechanisms underlying decoding signals

Recent studies have demonstrated the ability of classification algorithms applied to fMRI data to decode visual stimuli. Surprisingly, these algorithms decoded information segregated in cortical columns, e.g. ocular dominance and orientation, although the voxel size was large (3×3×3 mm3) relative to the width of columns (1 mm or less) in humans. The mechanism by which low-resolution imaging decodes information represented at higher resolution is not clear. Biased sampling of columns by the large voxels has been hypothesized. Alternatively, draining regions of blood vessels (BV) that cover functional regions non-homogenously could cause selective responses of BV, which might be captured by the large voxels. This study aimed at testing the two hypotheses on the mechanism underlying selective signals at low-resolution, and at comparing the decoding performances obtained at low- and high-resolution. High-resolution (0.5 mm) Gradient Echo (GE) and Spin Echo (SE) fMRI data were obtained from 3 subjects. Each scan included an epoch of baseline, and alternating epochs of left- or right-eye stimulation. Ocular dominance columns (ODC) were imaged in one slice overlapping the upper or lower bank of the calcarine sulcus. A multi-variate support-vector machine algorithm was applied to single fMRI images from the average scan at 3 different spatial resolutions. Voxels were sorted according to a rank that characterized them as belonging to gray matter (GM) or BV regions. The relative information for decoding the stimulated eye conveyed by signals sampled from ODC within GM and from BV was computed. Decreases in the effective contrast between ODC and in the decoding success rate were observed with decreasing resolution. At the lowest resolution (2 mm), the success rate was slightly above chance level when using data from < 0.1 cm3 volume of cortex. In contrast, high correct decoding rates of >0.8 for SE and GE data were achieved using even a small volume (less than 1 mm3) obtained at resolution of 0.5 mm. Whereas GM regions were slightly more informative on average, voxels overlapping with BV did carry information on the stimulated eye. In summary, using fMRI GE signals, the mechanism underlying the decoding signals involves contributions from both GM and macroscopic BV. We hypothesize that draining regions biased towards ODC with preference to one eye underlie the stimulated eye specificity of BV. Given that BOLD signal at 3T includes contributions from blood, we expect even a larger contribution from macroscopic BV to decoding signals at that magnetic field compared to 7T. Decoding at high-resolution is superior to low-resolution when applied to data from small cortical volumes