Generate interleaved DWI synthetic data

[valosekj]

We discussed that having of synthetic DWI data would be nice for performance testing of sct_dmri_moco with -interleaved option.

Possible solution:

• take some existing DWI data acquired by NON-interleaved acquisition (e.g. here)?
• apply random TxTy slicewise shift with regularization on every DWI volume to simulate interleaved acquisition

[jcohenadad]

I would go with pure synthetic dataset. E.g., create a multiple volumes with dummy segmentation (code here). This segmentation would be slightly modified (by modifying scalars to polynomial function defining the centerline) factor changed) across volumes, as to simulate non-linear motion (slicewise regularized thanks to the polynomial function). Having the ground truth, it is possible to assess the quality of registration.

Updated 2020-04-28 16:04:04: Actually, we would need to update dummy_segmentation() with the code here to allow polynomial function.

[valosekj]

Thanks for feedback. I am digging into it.