Open dataset OASIS-3 to BIDS

Hello everyone!

I am using the open dataset called OASIS-3. I would like to process it using fMRIPrep toolbox, but it works only with BIDS validated dataset. The data for one patient in OASIS dataset is organized like this:

+---anat2
|   +---BIDS
|   |       sub-OAS30457_ses-d0102_run-01_T1w.json
|   |       
|   \---NIFTI
|           sub-OAS30457_ses-d0102_run-01_T1w.nii.gz
|           
+---anat3
|   +---BIDS
|   |       sub-OAS30457_ses-d0102_run-02_T1w.json
|   |       
|   \---NIFTI
|           sub-OAS30457_ses-d0102_run-02_T1w.nii.gz
|           
+---BIDS
|       dataset_description.json
|       
+---dwi1
|   +---BIDS
|   |       sub-OAS30457_ses-d0102_dwi.bval
|   |       sub-OAS30457_ses-d0102_dwi.bvec
|   |       sub-OAS30457_ses-d0102_dwi.json
|   |       
|   \---NIFTI
|           sub-OAS30457_ses-d0102_dwi.nii.gz
|           
+---func1
|   +---BIDS
|   |       sub-OAS30457_ses-d0102_task-rest_run-01_bold.json
|   |       
|   \---NIFTI
|           sub-OAS30457_ses-d0102_task-rest_run-01_bold.nii.gz
|           
+---func2
|   +---BIDS
|   |       sub-OAS30457_ses-d0102_task-rest_run-02_bold.json
|   |       
|   \---NIFTI
|           sub-OAS30457_ses-d0102_task-rest_run-02_bold.nii.gz
|  

In terms of BIDS specification the filenames are fine, I think. However, the general structure of folders and sub-folders are not really in BIDS format. I have checked the dataset_description.json file and it says (more details are provided in pdf file here):

{
“HowToAcknowledge”: “Refer to documentation at www.oasis-brains.org”,
“License”: “Refer to Data Use Agreement at www.oasis-brains.org”,
"BidsVersion": “1.0.1”,
“Name”: “OASIS3”,
“DatasetDOI”: “https://central.xnat.org/data/archive/projects/OASIS3/subjects/CENTRAL_S05499/experiments/CENTRAL_E09988
}

I am a little bit confused here, but I think that the dataset probably in the older version of BIDS which is really different from the latest one. That is why it looks like the BIDS dataset, but not exactly - because of the version differences.

Since I am not so experienced user of BIDS, I see two options (1) write a script which can re-organize the folders structure to make it validated for BIDS, but it is can be very time consuming; (2) “upgrade” the dataset to newer BIDS version using ready solutions, but I am not sure that such things exist.

Do you have any ideas?

Any information and comments are useful :slight_smile:

More accurate searching in OASIS-3 documentation gives me the script to download data in BIDS format. You can find it here - download_scans/download_oasis_scans_bids.sh section.

But during BIDS validation I have a couple of warnings:

1: [WARN] You should define ‘SliceTiming’ for this file. If you don’t provide this information slice time correction will not be possible. (code: 13 - SLICE_TIMING_NOT_DEFINED)

2: [WARN] The recommended file /README is missing. See Section 03 (Modality agnostic files) of the BIDS specification. (code: 101 - README_FILE_MISSING)

The SliceTiming info should be in JSON file near the _bold.nii.gz file. I have JSON files, but the parameter is missing. Do you have any ideas why? Is it possible to extract this from the data and append to JSON file?

I am attaching a JSON file:
sub-OAS30027_ses-d0433_task-rest_run-01_bold.txt (1.8 KB)

Hi @egor.levchenko

Thank you for your message! Perhaps to clarify - the dataset is BIDS validated if no errors show up in the validation. These messages are warnings. These are good to be aware of but the dataset is valid. Typically, the slice timings are found in the dicom files and put together during conversion. For the cases they aren’t, determing the slice timings are much more challenging. In the preprint the authors mention the slice timings should be present in the accompanying json file. Perhaps reaching out to the authors?

Thank you,
Franklin

Hi @franklin,

Thank you for your response! Yes, I know that the dataset is valid, but just thinking to fulfil blank spaces to feel more confident about the preprocessing and flexible if ‘SliceTiming’ would be super important at some stage.

I asked the authors of the preprint and slice timing is available for many scans but not bold scans collected with software “syngo_MR_B13_4VB13A”. They working now to solve it and probably in future this problem wouldn’t exist anymore.

BW,
Egor

1 Like