Pseudonymise your dataset
If you plan to share your dataset then it is important to pseudonymise/anonymise your dataset in line with data protection laws. You should ensure JSON sidecar files do not include identifiable information and that anatomical images have facial structures removed.
It is best to store pseudonymised data in a separate dataset from the source identifiable data. The reason for this is if you are using version control such as DataLad then it might be possible to retrieve identifiable information from the version history.
Removing facial structures with PyDeface
PyDeface is a BIDS app that allows easy removal of facial structures from anatomical MRI scans.
Install PyDeface
Install with:
pip install pydeface
How to use PyDeface
Assuming you will be storing pseudonymised data in a new dataset, you can use the following command:
pydeface \
--outfile pseudonymised_dataset/path/to/T1w.nii.gz \
--force \
sourcedata/path/to/T1w.nii.gz
The --outfile argument specifies the output path. The --force argument will overwrite
the output if it already exists. It is useful to leave this on to prevent errors. Finally,
the path to the original anatomic image is the only positional argument required.
Pseudonymisation with BIDSonym
Todo
Add info re: BIDSonym