Problems with newer Siemens 4D Twist

[tpenzkofer]

Describe the bug

With newer Siemens MRI twist sequences the resulting 4D nifti is not ordered correctly (when scrolling through the timepoints, it scrolls slightly down and rotates quickly through the phases). Older Twist sequences convert fine.

see https://user-images.githubusercontent.com/1944948/224773051-6bbf5186-6fbb-4ee8-954e-c9867e9b6645.mp4

To reproduce

Steps to reproduce the behavior:

1. convert using e.g. "dcm2niix.exe -f "%s_%e_%d" -m y -i n -b y -z i -x i -o "
2. view using nii viewer

Expected behavior

A working 4D dicom which scrolls through time not space.

Output log

Convert 1300 DICOM as test_nii4\10_1_t1_twist_tra_dyn (192x192x26x50)
Ignoring derived image(s) of series 7 test\0\MR.1.3.12.2.1107.5.2.19.45227.30000022121613301282900000224
Warning: Siemens MoCo? Bogus slice timing (range -1..32800, TR=820 seconds)
Warning: Siemens XA exported as classic not enhanced DICOM (issue 236)

Version

Chris Rorden's dcm2niiX version v1.0.20211006 (JP2:OpenJPEG) (JP-LS:CharLS) MSC1900 (64-bit Windows)

Troubleshooting

Tried newest Version, no effect. Also tried different NII viewers, same problem. Tried different settings for -x (y,n,i)

[neurolabusc]

I note that you are using v1.0.20211006. The Siemens XA data has evolved a lot since 2021. I would suggest you first test the current stable release (v1.0.20220720). If this fails, you should try the development branch (currently v1.0.20230311) - you can compile this yourself or bet a pre-compiled version by going to AppVeyor, clicking on your OS (win for Windows) and then downloading the artifacts.

The only reasons that I have not made a new stable release is that I am working with Siemens engineers to understand some unexpected values in XA datasets. While XA cleans up DICOM a lot, there are still some rough edges. I would encourage you to work with the Siemens Research Collaboration Manager associated with your center to ensure that you are acquiring archival quality datasets.

[tpenzkofer]

Thank you for the quick answer, I tried with all versions up to the current development branch (downloaded the artifact) - same result - sorry if i hadn't made that clear

[neurolabusc]

I am unfamiliar with this sequence and can provide no more insight unless you can share an example datasets with my institutional email. The role of your Siemens RCM is to help you with precisely these sorts of situations.

[neurolabusc]

I would heed dcm2niix's warning Siemens XA exported as classic not enhanced DICOM referencing issue 236. Be aware that the XA classic DICOMs are impoverished, and that limits the capabilities of any existing or future tool. For archival quality XA data, you really want to export as enhanced data. Does exporting the data as enhanced DICOM resolve your problem?

[tpenzkofer]

Thank you for the helpful comments, will investiage further. Please note that the files were not exported at the console but pulled from PACS.

I can confirm that the images working are not X30 images and the ones with the problems from the X30 platform

[neurolabusc]

@tpenzkofer the latest commit to the development branch should resolve your issue. This issue is specific to the Siemens fldyn3d1 and requires that XA* images do not use the kludge described in issue 252 for VE*. Now the volumes are distinguished by acquisition number (0020,0012) rather than series number (0020,0011). This reflects an improvement in Siemens DICOM images across generations.

[tpenzkofer]

It works, thank you for that extraordinarily swift response!