Open villalonreina opened 3 years ago
Hey @villalonreina,
When you create the dmipy gradient table you can indicate what is the b-value threshold for counting a measurement as a b0. You can avoid this error in the future this way :-)
Let me know if it solves the problem!
Thanks @rutgerfick. Yes, we have this line in the code already:
scheme = acquisition_scheme_from_bvalues(bvals*1e6, bvecs, small_delta,
big_delta, b0_threshold=50e6,
min_b_shell_distance=50e6)
Our b-zeros are actually b=5, and the code still crashes, unless we modify the bval file. I tracked the error and it took me to the dmipy/signal_models/cylinder_models.py
as noted above.
You're right this is a bug.
merging the fix tomorrow morning, it's really small.
Hey @villalonreina , please update to the latest version of dmipy and let me know if this solves this particular problem.
Hi @matteofrigo @rutgerfick We have been trying to run MCMDI models on UKBB data and we were getting the following error:
After analyzing the error, we found out that the b0s in UKBB are not full b0s, but instead they are b=5.
Hence, there is an issue in
dmipy/signal_models/cylinder_models.py
because of the way it indexes bvals greater than zero (bval > 0). See below: