Can the non-human primate core-belt model be applied to the human auditory cortex? Evidence from functional and structural MRI at 7 Tesla
Julien Besle (1), Olivier Mougin (2), Rosa Sanchez-Panchuelo (2), Penny Gowland (2), Richard Bowtell (2), Sue Francis (2) & Katrin Krumbholz (1)
(1) MRC Institute of Hearing Research, Nottingham, UK
(2) Sir Peter Mansfield Magnetic Resonance Centre, University of Nottingham, UK
Anatomical and electrophysiological studies in non-human primates (NHP) have subdivided the auditory cortex (AC) into core and belt areas showing distinct structural and functional properties. It remains unclear however whether analogous subdivisions exist in the human AC and how they are laid out on the supratemporal plane. Post-mortem and in-vivo anatomical evidence shows that a core area with similar properties as in the NHP is located along Heschl’s gyrus (HG), whereas tonotopic mapping using fMRI suggests that the main axis of the core and belt areas should cross HG. Recent models of the human AC aiming at reconciling these data with the NHP model have proposed that core areas might in fact be at an angle with HG and that tonotopic gradients in difference sub-areas of the core and belt might not be aligned. Here we measured tonotopy, frequency selectivity and myelination in 12 subjects at ultra-high magnetic field (7T) and specifically tested the recently proposed models.
For the functional mapping (sparse 2D GRE EPI, 1.5 mm resolution, phase-corrected for B0-related distortions), we estimated best frequency and frequency selectivity using trains of narrowband noises at 7 single centre frequencies. For structural mapping of myelination, we estimated the R1 longitudinal relaxation rate (3D MP2RAGE, 0.6 mm resolution) and the magnetization transfer ratio (3D MTRAGE, 0.7 mm resolution). All structural and functional measures were projected onto a flattened model of the supra-temporal cortex, segmented from the high-resolution processed MP2RAGE volume. Structural measures were corrected for cortical thickness and curvature. Group-averaged maps were created using spherical registration.
In all subjects/hemispheres, we identified 2 central tonotopic gradients (high-to-low-to-high) oriented, each oriented at 70° relative to HG (and 140° with each other), confirming that the main tonotopic gradients are oriented across HG, although they are not exactly aligned. As in previous reports, there was also a posterior high-low gradient on the planum temporale. In most hemispheres, a core area of high myelination and high selectivity was found aligned with HG (only in its medial part for myelination), with no evidence for an angle between core and HG. On the group-averaged maps, the area of higher myelination/selectivity corresponded mostly to the most anterior central gradient that was previously identified with R in the NHP model. In individual maps however there was considerable inter-subject variability, with higher myelination and selectivity corresponding to either or both of the central gradients, although it overlapped more often with the anterior than the posterior gradient. Overall, these results suggest that even a tweaked version of the NHP model cannot be applied to the human auditory cortex.