This forum seems the correct location for recent questions on neurostars and nitrc regarding NIfTI to DICOM conversion. Slicer is the tool I would use for these operations. Perhaps members of this forum can provide guidance to these users.
Alex Cohen raises an interesting question of whether the rich meta data of the BIDS JSON headers could be included in the NIfTI to DICOM conversion. This would have several benefits:
- DICOM images can use a lot of tools that do not support NIfTI.
- Complicated 2D classic DICOMs converted to BIDS/NIfTI could be saved as vendor agnostic simple Enhanced DICOM resulting in faster transmission and less disk space.
- Vendor specific private tags (e.g. for diffusion gradients GE uses different frame of reference than Siemens/Philips) could be saved to common public tags (bearing in mind NIfTI uses yet a different frame of reference).
- Privacy issues that have plagued DICOM due to buffer-overflow and failure to zero-pad arrays could be addressed. A simple lean converter developed with security in mind could prevent data leakage.
- Users could edit the JSON text files to insert crucial data not provided by the vendor and these would then be inserted in DICOM files (e.g. Philips DICOM does not yet report slice timing or phase encoding polarity).
I realize there are some challenges to this approach:
- There are no public tags for some details (slice timing, phase encoding polarity, etc).
- Since different vendors have interpreted the DICOM standard differently, many tools use vendor specific heuristics and therefore may not work well (at least initially) with vendor agnostic data.