Hi all,
Does anyone know of a reliable method for converting a 4D BOLD image from NIFTI format to DICOM?
I’ve tried out a number of resources/softwares to some extent, but have not found one yet that preserves data quality, voxel values, and bit resolution.
Here’s a summary of what I’ve tried so far:
nifti2dicom - I have not figured out how to reliably handle MR-modality images; documentation is limited and the program is not returning any feedback on why the command is failing/not generating anything.
nii2dcm - Does not support 4D images. Tried using fslsplit and then running basic command on each volume, but bit resolution was truncated - no negative or floating point voxel values are retained, and the range is forced to [0,4168], when the original nifti’s values are in the range [-140,140]. Not sure if there’s a way to change this.
3D Slicer - Closest to the original nifti out of all the methods I have tried so far. I am able to load in each volume as a nifti and export as dicom, but the floating point values are not retained when exported as short. The dicom output options are UnsignedChar, Char, UnsignedShort, Short, UnsignedInt, and Int.
I am erring to the preference of the 3D Slicer method, but I am concerned about the truncation of the floating point values for the sake of our outputs.
For each of these methods, I am also getting weird outcomes as far as image origin and orientation go, but I figure these are more easily remedied post-hoc than the major data value issues, so I would like to prioritize the data values first.
Any recommendations or suggestions to improve these methods are greatly appreciated. My PI has a collaborator who will only accept our processed images in dicom format, so I need to figure out the most accurate way to convert and share this data as possible.
Thanks in advance,
Jess