Why can't we just use the -b of dcm2niix instead of using dcm2bids, bidsify, HeuDiConv, etc.?

The -b option of dcm2niix seems to cause its output to be in BIDS format.

So why do we have other tools that convert from DICOM/nifti to BIDS? Doesn’t dcm2niix already do that?

Is there some difference between the output from dcm2niix called with the -b flag and the output from tools like dcm2bids, bidsify, HeuDiConv, bidskit, dac2bids, etc.?

dcm2niix will convert DICOM images to BIDS-format NIfTI images with JSON sidecars. However, it will not give these files the correct filename or put them in the correct folder. BIDS requires specific filenames (‘sub01-T1w.nii.gz’) in specific folders (/sub01/anat). Since dcm2niix does not know your intention (e.g. subject number, usage of scan sequence), it simply converts the DICOM data to NIfTI. Wrappers like dcm2bids, bidsify, HeuDiConv, bidskit, dac2bids, etc. combine these images with information regarding your intention to name the files as required by the standard.


Ah I see, that makes sense, thank you! One more question: you specified “BIDS-format NIfTI images”, which implies that there are also non-BIDS-format NIfTI images. Is that true? I thought that BIDS only dictated file names and structure, not the actual contents of the files.

BIDS makes several strong assumptions about NIfTI images. In general, fMRI and DWI datasets are saved as a single 4D file (as expected by FSL and AFNI) rather than a series of 3D images (optional but common for SPM). On the other hand, BIDS requires that multi-echo data MUST be split into one file per echo. dcm2niix conforms to these expectations - stacking single echo fMRI and DWI data as 4D images, separating multi-echo files into different NIfTI files. Further, some vendors save derived images (e.g. isotropic averages) in the same series as the raw DWI images, and BIDS expects those derived images to be removed. One can always use a tool like fslmerge and fslsplit to change the resulting images, but by default dcm2niix provides NIfTI images grouped as expected by BIDS.


I see, that’s extremely informative, thank you again for the help!