Abstract
Purpose: Primary visual cortex is one of the best studied areas of cortex. Since 1996 human visual cortex can be mapped using functional MRI in vivo. Up to now these measurements have provided a visualization of the retinotopic architecture only for individual human subjects. To characterize human visual cortex across individuals, we developed a new method to obtain an atlas averaged across many brains.
Methods: Retinotopic mapping data was collected using advanced fMRI procedures. We obtained reliable maps down to 0.37 deg eccentricity using log-scaled eccentricity mapping stimuli and an extended fixation cross that optimized the stability of fixation. An improved version of the atlas fitting functions from the VISTA-toolbox (Dougherty et al., 2003 J.Vis. 3:586–598) was used to enforce continuity of retinotopic measurements. A grid of 10×12 locations was identified in each subject and the distance between each pair of these locations was measured along the reconstructed 3D manifold. and reconstructed with Multidimensional Scaling of the distance grid averaged across subjects.
Results. We find an increase in V1 width up to 8 eccentricity (by a factor of 3.1, from 17.4 3.4 mm at 0.37 to 54.5 2.8 mm at 8, with no significant increase thereafter). Our data demonstrate that the shape of the 2-dimensional projection of V1 is closer to a rhomboid than to an elongated ellipsoid as generally reported. We fitted the mapping function w=log(z+a) and estimated an optimal fit with a=0.78 (as proposed by Schwartz, 1993), however the model failed to describe important properties of our reconstruction, especially in the parafoveal area from 0.5 to 2. We also compared more elaborated models such as the conformal dipole mapping (Balasubramanian et al. 2002 Neural Netw. 15:1157–1163).
Conclusion: These data also provide a detailed framework for determining magnification functions and anisotropy in human visual cortex.