 |
 |
Dear experts, I have two questions regarding bedpostx, one about the nfiber option, and the other about GPU use. Regarding the nfiber argument, I was under the impression that this number should be limited to 1 or 2 for a non-HARDI sequence (say 30 directions) but this Neuroimage paper (doi.org/10.1016/j.neuroimage.2014.12.060) seems to find very good results (bedpostx top performer) when using nfiber=3 even with a low number of directions. Would you confirm that using nfiber=3 even for diffusion MRI sequences with 20/30 gradient directions is fine? Regarding GPU use, I was wondering what is the recommended method to indicate bedpostx_gpu which GPU id to use (for users with at least 2 GPUs and wanting to use a specific one). Best wishes, Michael