dcm-zurich BIDSification

[valosekj]

Overview

This PR discusses the BIDSification of dcm-zurich dataset.

Sample of the input dataset (located on joplin in ~/duke/projects/dcm-normalization/data_zurich_2023-02-22/Nifti_proCSM/Nifti_proCSM):

 ├── 250791
 │	├── t2_tse_sag_384_25mm_0005
 │	│	└── s250791-113456-00001-00020-1.nii
 │	├── t2_tse_tra_oben_0006
 │	│	└── s250791-113720-00001-00015-1.nii
 │	└── t2_tse_tra_unten_0007
 │	    └── s250791-113952-00001-00015-1.nii
 ...

Sample of the output BIDS dataset (located on joplin in ~/extrassd1/janvalosek/dcm-zurich):

 ├── code
 │	└── curate_dcm-zurich.py
 ├── dataset_description.json
 ├── README.md
 ├── participants.json
 ├── participants.tsv
 ├── sourcedata
 │	 ├── dataset_description.json
 │	 ├── sub-250791
 │	 │	 └── anat
 │	 │	     ├── sub-250791_acq-axialBottom_T2w.json
 │	 │	     ├── sub-250791_acq-axialBottom_T2w.nii.gz            # non-stitched image
 │	 │	     ├── sub-250791_acq-axialTop_T2w.json
 │	 │	     └── sub-250791_acq-axialTop_T2w.nii.gz               # non-stitched image
 ...
 ├── sub-250791
 │	 └── anat
 │	     ├── sub-250791_acq-axial_T2w.json
 │	     ├── sub-250791_acq-axial_T2w.nii.gz                          # stitched image
 │	     ├── sub-250791_acq-sagittal_T2w.json
 │	     └── sub-250791_acq-sagittal_T2w.nii.gz
...

README.md is provided within this PR to allow easy review.

Stitching

Most subjects (n=121) have two T2tra images. One covers the upper cervical SC (top FOV), second covers the lower cervical SC (bottom FOV). I stitched these two images using sct_match -stitch:

data-management/scripts/curate_dcm-zurich.py

Lines 278 to 279 in 0c01dac

	os.system('sct_image -i ' + path_file_top + ' ' + path_file_bottom + ' -stitch -o ' + path_file_stitched +
	' -qc ' + path_qc + ' -qc-subject sub-' + subject)

I also stored the original non-stitched images under sourcedata folder (point 2 of "storage of derived datasets").

I checked the QC (~/extrassd1/janvalosek/dcm-zurich/qc) for 121 subjects where the stitching was applied, and it looks fine.

Checks

bids-validator is reasonably happy. The upload of the dataset to git-annex will be discussed in a separate issue, as usual.

Below are some checks that the number of subjects between the input and output datasets is the same.

The number of subjects in the input dataset:

$ cd ~/duke/projects/dcm-normalization/data_zurich_2023-02-22/Nifti_proCSM/Nifti_proCSM
$ ls -1d * | wc -l
140

The number of subjects in the output dataset:

$ cd extrassd1/janvalosek/dcm-zurich
$ ls -1d sub-* | wc -l
140

The number of subjects in the output dataset under sourcedata (i.e. subjects with two T2w axial images --> stitching was done, and the original non-stitched images are stored under sourcedata):

$ ls -1d sourcedata/sub-* | wc -l
121

There are also:

• 17 subjects with only a single T2w image --> no stitching was performed (see README.md)
• 2 subjects with no T2w image at all (see README.md)

121 + 17 + 2 = 140 (i.e. the number of subjects is the same between the input and the output datasets)

[valosekj]

@jcohenadad Could you please review if README.md is fine?

[jcohenadad]

Awesome! thanks

[sandrinebedard]

Everything looks good!

[valosekj]

Thank you for your feedback!
The dataset has been pushed to git-annex within #218 --> closing this PR