dMRI gradient data implementation is inconsistent

[jhlegarreta]

What happened?

When attempting to run the tests in PR #313, it becomes apparent that the gradient data format is not consistent across the code. The data.filtering.dwi_select_shells functionexpects an Nx4 array, the b-values being the last column in:

nifreeze/src/nifreeze/data/filtering.py

Line 257 in e92e725

bvalues = gradients[:, -1]

However, the rest of the code handles a 4xN array, the b-values being the last row. Note that the transpose of these arrays does not give the other array.

This is not captured by any test.

This becomes apparent when debugging the tests in PR #313. For the avg model and DTI model tests to pass,

nifreeze/src/nifreeze/data/filtering.py

Line 257 in e92e725

bvalues = gradients[:, -1]

needs to be changed to

bvalues = gradients[-1, :]

What command did you use?

pytest test/test_integration.py

What version of the software are you running?

3139500

How are you running this software?

Local installation ("bare-metal")

Is your data BIDS valid?

Yes

Are you reusing any previously computed results?

No

Please copy and paste any relevant log output.

if not shellmask.sum():
>           raise RuntimeError(f"Shell corresponding to index {index} (b={bcenter}) is empty.")
E           RuntimeError: Shell corresponding to index 3 (b=1000.0) is empty.

Additional information / screenshots

No response

[arokem]

Following discussion with @jhlegarreta and @oesteban : we agree that the implementation should be consistently N x 4, rather than the current implementation. However, this entails refactoring the existing code in https://github.com/nipreps/nifreeze/blob/main/src/nifreeze/data/dmri.py#L127-L133 and fixing up everything that flows downstream from there. So, the code in filtering is correct, but everything else needs to be changed 😄