Hello Chris and the `dcm2niix` community,

I'm reaching out for guidance on a challenge we're currently facing. We are working with abdomen MR DICOM images intended for segmentation model training. We've observed that some of our DICOM datasets, which are composed images (resulting from upper and lower acquisition sequences combined together), exhibit inconsistent actual slice spacing. I understand that the NIfTI format typically assumes equidistant slices.

Having not tried the `dcm2niix` conversion for these particular datasets yet, I'd like to be proactive and gather insights:

1. Is it possible to convert such DICOM images with varying slice spacings to NIfTI using `dcm2niix` without running into issues?
2. If challenges are anticipated, do you have any recommendations or best practices for preprocessing these DICOM images prior to conversion?
3. Are there specific flags or parameters in `dcm2niix` that might help in ensuring an accurate conversion for these datasets?

Thank you for your time and expertise. We truly appreciate any insights you can provide.

Chenglin

It sounds like your scans will be composed of several different series, each with different resolutions and FOVs. dcm2niix converts each DICOM series as a separate NIfTI file. Therefore, you will have one NIfTI image per series. You will then want to use a different tool to concatenate them all together as a single file. The NIfTI format requires that all voxels are equidistant, so some interpolation will need to be required if you have different FOVs or distance between slice centers. The task of stitching different series together is outside the remit of dcm2niix, so you will have to use a different tool.