Wrapper dev - attribute harmonization and signal normalization #106
Conversation
src/wrappers/OsipiBase.py
Outdated
| args = [data[ijk], use_bvalues] | ||
| # Normalize array | ||
| single_voxel_data = data[ijk] | ||
| single_voxel_data_s0 = single_voxel_data[0] |
There was a problem hiding this comment.
This is assuming that the first index will be the b=0 data. Probably valid but is that an requirement we want to make? Seems like something someone is going to get caught on at some point.
I actually have an ancient PR that should find the b=0 automatically here: https://github.com/OSIPI/TF2.4_IVIM-MRI_CodeCollection/pull/80/files#diff-3e73b3e25749003f8536e5b52b164fa715d358ea4237220c3a1e32f445be03a7
Would dusting that off be good to include here?
There was a problem hiding this comment.
I agree with having something more sophisticated than this. I remember raising this and it sort of barely landed on us just requiring it to be first. But I like the solution of taking the mean of all b0's in this example, and shouldn't be hard to implement as we have all the b-values in an array. So I'll adjust the code to include this functionality.
Yes, since the attributes are included in a figure Oscar generates. Also the signal normalization is a feature we sort of promise in the text, and is needed for the bounds and initial guesses to also be harmonized. |
…ue signals and normalized the entire array to its mean.
|
Looks good to me! Happy to merge |
Describe the changes you have made in this PR
Two things in this wrapper update:
Checklist