Add anatomical data for testing

[po09i]

Anatomical data is required to start implementing masking functions. The data could be unsorted dicom data so that we can also improve our bids config file.

[jcohenadad]

The data could be unsorted dicom data so that we can also improve our bids config file.

you need nifti input for unit-testing your masking function, not dicom

also, i would suggest to separate testing as much as possible, i.e.:

• to test the dicom_to_nifti function: add under unsorted dicom the necessary files (no more, no less than required-- trying to keep the repos as small as possible)
• to test masking: add nifti file in the already-existing BIDS folder

[po09i]

I'm thinking of adding the acquisition done in ACDC_108:

• anat: gre_realtime_zshim -- size (68, 128, 20, 8)

Your suggestion is to add an unsorted dicom version of this acquisition and a nifti bids version to facilitate unit tests?

[jcohenadad]

my suggestion was to separate the tests data when it makes sense.

example:

• test 1 minimally requires data A
• test 2 minimally requires data A or B
• test 3 minimally requires data C

with that in mind:

• uploading A B C D E F is useless
• uploading A B C is suboptimal
• uploading A C is optimal

it sounds trivial, but it is important to always keep that in mind and try to minimize test data size.

example: you're planning on uploading a dataset with 8 echoes, but in fact you only need two so you could do:

• testing of fieldmap with another sequence with two echoes (because the 6-echo case is already tested with another already-existing data)
• testing of masking, which only requires one echo

so, why do you need to upload (68, 128, 20, 8)? you could instead get rid of echoes 3->8 so you save 68x128x20x6xdtype space.

the same applies to the XYZ matrix: do you absolutely need the full matrix? can the tests run with equal sensitivity with e.g. 32x32x5 matrix instead?

i think you get my point

[po09i]

Yes this makes things clearer. I'll wait until I start working on the masking function so that I can pinpoint exactly what is required for testing.

[jcohenadad]

I'll wait until I start working on the masking function so that I can pinpoint exactly what is required for testing.

and in fact, you can also generate data on the fly, e.g. https://github.com/neuropoly/spinalcordtoolbox/blob/master/spinalcordtoolbox/testing/create_test_data.py#L120