Curation of data received

[plbenveniste]

A first batch of data was received yesterday. It is only a sample to start creating the pipeline. It is in the DICOM format.

• Should this data be git-annexed ?
• If so, which git-annex dataset ?

This issue will also be used to describe the contrasts contained in the batch of data received.

[jcohenadad]

Should this data be git-annexed ?

usually, we would not do it for only one subject, too much trouble. On the other hand, anticipating that we will receive more data from this site (approx 90 patients), maybe it is in fact a good idea to create a git-annex entry for this dataset.

[plbenveniste]

As mentioned in #2 (comment), we only focus on focus on specific contrasts. Here is the code used to bidsify these specific volumes from DICOM to BIDS.

Code used

# For Csp Ax T2
dcm2niix -f 'sub-HC001_acq-axCerv_T2w' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/Csp_Ax_T2-20220509

# For Csp Sag 3D Fiesta
dcm2niix -f 'sub-HC001_acq-sagCerv_FIESTA' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/Csp_Sag_3D_Fiesta-20220509

# For Csp Sag Cube T2 Focus
dcm2niix -f 'sub-HC001_acq-sagCervCube_T2w' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/Csp_Sag_Cube_T2_Focus-20220509   

# For ORIG SAG T2 LOC UPPER DL
dcm2niix -f 'sub-HC001_acq-sagUpperLocalizer_T2w' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/ORIG_SAG_T2_LOC_UPPER_DL-20220509

# For Denoised SAG T2 LOC UPPER DL
dcm2niix -f 'sub-HC001_acq-sagUpperLocalizerDen_T2w' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/ORIG_SAG_T2_LOC_UPPER_DL-20220509

# For Csp Sag T2 STIR (only Csp)
dcm2niix -f 'sub-HC001_acq-sagCerv_STIR' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/ORIG_Csp_Sag_T2_STIR-20220509

# For Csp Ax MERGE (only Csp)
dcm2niix -f 'sub-HC001_acq-axCerv_MERGE' -o ms-mayo-bids/sub-HC001/anat -z y Anonymous_71e0421d-a61cfbc81cf9c8bbb0605733d6c68c5b1338b367/MR_INNOVATION-20220509/Csp_Ax_MERGE-20220509 

The data was moved to duke/temp/plben/ms-critical-lesion/ms-mayo-bids

[plbenveniste]

Full dataset was received and is stored in ~/duke/temp/plben/ms-critical-lesion/MRI_Mayo_Clinic_for_Montreal.zip.

TODO:

• Bidsification of the data in the git annex repo

[plbenveniste]

@mguaypaq Could you create a new git-annex dataset where I could store this data?

Name: ms-mayo-critical-lesions

[plbenveniste]

Data was received contains .nii.gz files → no need to perform dcm2bids !
There are 482 files → 241 images and their corresponding segmentation.
For each file, a label file was included. This file is the segmentation of the critical lesion.
The file names are not explicit. I have asked Dan for the subject ID mapping.

Lesion segmentations
However, it appears that the segmentation is not binary. Here are a few examples:

For image: 1.2.752.24.3.476272775.485368.1425075.nii.gz

For image: 1.2.840.113619.2.5.1762533628.1467.1016541915.458.nii.gz

For image: 1.2.840.113713.400.2.30354407.nii.gz

I suppose from this that the segmentations correspond to:

• the spinal cord segmentation on the axial slice of the lesion (segmentation value=2)
• the lesion segmentation on the axial slice (segmentation value=1)

Also, from my few observations, the lesions were segmented on only 1 slice (supposedly the central slice).

Corrupted files
Some files are corrupted:

Image: 1.2.840.113619.2.244.3596.13616250.25529.1410548851.731.nii.gz

Next steps:

• Perform subject ID mapping
• QC all images to identify other corrupted images
• Send an email to the team to explicit the process of lesion segmentation
• Bidsify the dataset
• Push the dataset to git-annex

[mguaypaq]

@plbenveniste here's the new repo, I've given you write access:
https://data.neuro.polymtl.ca/datasets/ms-mayo-critical-lesions

[plbenveniste]

Thanks @mguaypaq 🥇

[plbenveniste]

Data was preprocessed and PR was created : PR 1

[plbenveniste]

Follow-up from 2025-07-22 meeting. We discovered that the data which we received wasn't the correct one. The segmentations were supposed to include full-volume segmentation of all MS lesions (not only critical lesions). They will send us the updated dataset.

[plbenveniste]

Email received on 2025-09-22:

Which MRI sequences were used for these acquisitions?
Ax T2 TSE, Axial T1 TSE post gad, Sag T1 TSE, Sag T2 TSE, Sag IR (cord w/FS) Inversion time 180

What field strength was employed?
3T was used in all the cases from Mayo

Were the images obtained using an inversion recovery sequence?
The Sag IR Acquisition is an inversion recovery sequence.

They are currently checking the data they sent us to include all acquisitions.