BIDS file names for lesion masks

We are about to release a lot of new data for our chronic Aphasia Recovery Cohort (ARC) on OpenNeuro, as well as a new acute data release. However, I am confused about how to report the lesion masks as based on these neurostars and Github

The core issues are:

  1. Lesions are in native space for a specific modality (e.g. aligned to T2w) which may have different dimensions and angulation from the T1
  2. These are not in standard space, so I am unclear if they should exist in the derivatives folder (they were not acquired by the scanner).
  3. Lesion expansion is common, and therefore the masks are unique to a single session.
  4. The corresponding BIDS sidecar will have the “Type: Lesion” field

Consider a participant that was seen twice at 1058 and 1492 days post stroke, is this the correct nomenclature for a lesion drawn to match the T2 scan acquired on the visit at 1058 days post stroke?

├── ...
├── sub-M2001
│     ├── ses-1058
│     │     ├── anat
│     │     │    ├── sub-M2001_ses-1058_T1w.nii
│     │     │    ├── sub-M2001_ses-1058_T2w.nii
│     │     │    └── sub-M2001_ses-1058_T2w_mask.nii
│     │     └── func
│     └── ses-1492
│           ├── anat
│           │    ├── sub-M2001_ses-1492_T1w.nii
│           │    ├── sub-M2001_ses-1492_T2w.nii
│           │    └── sub-M2001_ses-1492_T2w_mask.nii
│           └── func
└── ...

Hi Chris. I would probably name these masks in this way:


T2w is a non-standard space, so these would be expected to have the SpatialReference metadata. In this case, something like bids:raw:sub-M2001/ses-1058/anat/sub-M2001_ses-1058_T2w.nii would be appropriate, making use of BIDS-URIs.

That is correct to use the "Type": "Lesion" metadata.