Hi Mrtrix3 peeps,
I have been attempting to perform the the dwi2response msmt_5tt function on some data I collected. the data:
9 shells x 60 directions per shell
bvalues = 0, 100, 500, 900 1300 1700, 2000, 2300, 2700, 3000.
If I perform the dwi2response function using the full dataset, my white matter response function contains -nans as its first row. I also notice that it contains only 9 rows, and not 10. Is mrtrix somehow blending my b=0 and b=100 s/mm^2 data? If so, what can I do about this?
The NaN issue sounds similar to this one - and will hopefully have the same solution.
The merging of the b-values is trickier to address. Currently, volumes are grouped into shells if their b-values are within some tolerance of each other. Currently, this is hard-coded to 100 at this point in the code. The simplest fix is reduce that value to 50 or so. In the long run, we should make this a config file option - I’ll file an issue for that on GitHub.
Ok yes you are correct that was the issue. To test, I replaced the 100 with 101, and I was able to correctly perform the msmt_csd analysis! even the NaNs went away. I suppose that makes sense because trying to fit spherical harmonics to a shell with signal radii corresponding to both b=0 and b=100 at the same time would prove tough?