Mri
I only have nifti files and no dicom. Can I still create a BIDS dataset?
In theory yes, but it is possible that you will be missing some metadata that is required by the BIDS specification.
A couple of BIDS converters can work with nifti files:
- bidscoin
- bidsme
- explore asl
- data2bids
See the list of converters here.
However...
-
They may not work with the datatype you have (for example not sure that explore ASL can deal with
func
data). -
Many of them still expect that you have a json side car for each nifti (BIDSme for example - from the top of my head).
-
They may expect a certain input structure to work efficiently, so we may have to move files around.
Regarding the "missing" JSON files
Depending on the datatype you are dealing with this can be more of less annoying.
2 examples:
For the most typical anat
files,
they don't have any REQUIRED metadata,
so you could have just the data files without any accompanying JSON.
If you get into more exotic anat files (like for quantitative MRI)
then this may become a problem.
For func
files you only need TaskName
and RepetitionTime
and the former you can decide what it is and the latter should be in the Nifti header.
So you should be OKish there too.
{
"TaskName": "name_of your task",
"RepetitionTime": your repetition time value
}
Obviously you will be missing some metadata that would be required for some type of preprocessing (like slice timing info).
Tips
-
If you have the PDF with the details of acquisition sequence or a method section from a paper with that data, you could "recover" some extra metadata.
-
SPM dicom import tool usually leaves a couple of metadata in a description field of the nifti header. That can be a problem for data anonymisation but may help you in your case if this is the tool that was used. See an example here.
-
If you have to script tings manually rather than using a converter remember to use pybids path construction tools or the bids matlab equivalent to make your life easier when constructing bids valid filenames.
What defacing tools can I use?
If you want to share your BIDS data set, chances are that you will have to anonymize it and therefore prevent identification of the participants from their anatomical scans. You will need to de-identification (or deface) the anatomical images. There are several options to do that
- If you already have a valid BIDS data set you can simply use the BIDSonym BIDS app on it. To do that it relies on several tools that you can also use before you have finalized your BIDS data set
Otherwise you can also use:
- Fieldtrip under matlab can do it.
- SPM8 and SPM12: when in the batch editor go to:
SPM menu --> Util --> Deface
Generated by FAQtory