aDWI-BIDS - Diffusion Imaging Metadata for BIDS
James Gholam
Hello all,
This is a preliminary message concerning mine and my colleagues’ efforts over the last year in drafting a data standard to properly label diffusion MRI metadata (BIDS for advanced Diffusion Weighted Imaging, or aDWI-BIDS). We felt that BIDS offers substantial benefits to the neuroscience community, but lacked satisfactory support for the numerous variations in DWI experiments that may be run.
In our team’s prior work on the BIDS Toolbox (a dcm2niix / Bidskit interface), it became apparent that BIDS relied on a small number of fields to identify sequences. This led to overuse of the [_dir-<label>] field to describe parametric variation. Worse, in the diffusion community, it has become typical to use .bvec and .bval files in combination with other metadata to differentiate acquisitions captured with different parameters. This restricts heuristics to identifying the major type of a sequence, then assuming knowledge of its variation.
In diffusion imaging, variations in an experiment can significantly impact the interpretation of the data obtained, and how subsequent processing should be carried out. This limits the lifetime of often laboriously obtained data to the lifetime of the toolchain used to work with it. This is a long way from the modularity offered by other BIDS modalities. In our proposal, we seek to overcome these limitations with a means of generically communicating MR sequence activity over multiple experiments, with a means of tracking any parametric variation.
aDWI-BIDS is built within BIDS using tools already extant (namely a linked tsv / json structure), and uses an event structure within the JSON inspired by the HED. Support is also added for dense binary data for gradient / RF waveforms using references to external files in the json / tsv. Finally, we permit explicit labelling of parameter sets to regions of a NIfTI, permitting complex description of interleaved sequences.
aDWI-BIDS has been developed for some time privately by a group of collaborators from Cardiff, Lund, Harvard and Valladolid, alongside individuals from Siemens. We have withheld building a BEP until we were sure that the decisions made were well founded and tested. Having recently completed a first draft of the standard, we feel now is an ideal time to share this work with the wider BIDS community for critique and suggestions. Our standard is built for extensibility and community definition, and we hope that with your support that it will meet the needs of the DWI community within BIDS.
Our internal documentation will be compiled into a rigorous BEP over the next few weeks, and we hope that you will consider reviewing it as it grows.
Thank you all for building something worth nurturing,
James Gholam