mrconvert: Directions in 0018,9089 not detected

Hi all!

I’m trying to convert DICOM data from a preclinical scanner of the company Mediso. Conversion works in general, but while the b-value is detected correctly, the directions in the gradient table are all 0 0 0.

I can see in the DICOM data, that the directions are stored in 0018,9089. But this tag seems to be nested under 0018,9076 and FFFE,E000? Perhaps this nesting is the problem?

Output of dcminfo:

[DCM] 0018 9076 SQ       40     2258 > DiffusionGradientDirectionSequence
[DCM] FFFE E000 UN       32     2270   - Item
[DCM] 0018 9089 FD       24     2278     DiffusionGradientOrientation         0.93754 -0.34099 -0.068877

Your help would be very much appreciated!

Many thanks,

Hi Marco,

I’ve had a quick look into this, I think the issue is that the code in MRtrix expects this information to be stored in a multiframe DICOM, but it looks like this is a single-frame DICOM. I can probably fix this, but I’d be very concerned about breaking DICOM import in other contexts… If you can share some of these datasets, I could look into it and see if I can modify the code to support these types of data without breaking everything else… Would that be possible?


Thank you, Donald, for looking into this. Sure, happy to share a dataset. Will send you a download link via direct message.
Thanks again, Marco

Hi Marco,

Thanks for the data. It’s as I thought, and this requires quite a deep change to the way we handle DICOM (mostly conceptually – the change to the code itself is relatively minor). I’ve created a pull request for it, which you can find here. If you’re comfortable compiling from source, please check it out and have a go. I’m a bit hesitant to merge it to the main codebase too early given how critical correct DICOM handling is to everything, so it might take a while before this makes it out to release…

Hi Donald,
Just compiled the branch and can confirm that it works. Thanks a lot for adding this so quickly! :grinning:

1 Like