Home Computer Science Computational Diffusion MRI: MICCAI Workshop, Athens, Greece, October 2016
On the temporal lobe we note that both the myelin map and diffusion-based parcellation exhibited distinct patches that roughly coincide with primary auditory core and belt areas (BA 41 and 42). This suggests that some structural information in the dataset is maintained despite Heschl’s gyrus exhibiting a markedly variable folding pattern across subjects .
On the right hand side of Fig. 4 the posterior occipital lobe contained prominent purple, blue, and red clusters, giving it a distinct appearance compared to much of the rest of the medial cortical sheet, which was assigned predominantly to green and
Fig. 4 The medial view of the diffusion-based parcellation (right) and the group average myelin map (left)
white clusters. This generally correlates with the high myelination of this region, seen in the T1/T2 data at the left of Fig. 4; however it is worth noting that the region of apparent heavy myelination in V2 just below the tip of the “18” arrowhead projected further in the superior direction than the purple region in the diffusion- based result.
The black contours outline the inner and outer extents of area 18 from the FreeSurfer probabilistic atlas, i.e the secondary visual area V2. The inner boundary of this contour corresponds to the neighbouring primary visual cortex, V1 (area 17), within the calcarine sulcus. Despite V1 possessing a prominent tangential band in layer 4B that is lacking in extrastriate area V2 [5, 17], we did not observe distinct coherent clusters corresponding to the full extent of these two areas in either the T1/T2 data or the diffusion data. Instead the most salient feature of this region was the red cluster, which is located near the upper vertical meridian of V1. It is unclear why the boundary of V1 was characterised uniquely by the dMRI feature set, rather than the entire region. However, the myelin map showed a significant decrease in myelination in the same location that is not consistent with the underlying anatomy. This suggests the presence of a systematic surface placement error that my have resulted in CSF partial voluming in both data sets, which resulted in a region near the upper and lower vertical meridian border between V1 and V2 standing out.
Finally, returning to the lateral surface (see Fig. 2) in the middle of the myelin- dense region of MT+ it is possible to distinguish a border between a posterior orange cluster and an anterior white/tan cluster. A study examining the relation between quantitative T1 and retinotopy  surprisingly showed that the heavily myelinated oval in the lateral occipital cortex does not directly correspond to MT; instead, MT proper only accounts for the posterior part of that oval. The anterior part may correspond to FST, which represents parts of the visual field already mapped in MT, and which responds to the ipsilateral visual field unlike MT. Once again, this suggests that diffusion data may help distinguish regions not easily separated by using myelin density alone.
|< Prev||CONTENTS||Next >|