Siemens voxel oriented incorrectly with respect to structural MR

[PichardRarker]

Hi Suspect gurus.

I'm having trouble processing some Siemens Prisma data. If I use the following to co-register my MRS and T1-MRI data:

mrs = suspect.io.load_siemens_dicom("/path/to/mrs/dicom")
anat = suspect.image.load_nifti("/path/to/corresponding/t1.nii.gz")
voxel_mask = suspect.image.create_mask(mrs, anat)

... then the voxel is rotated 90 degrees compared to the true location (as confirmed using the voxel-T1 overlay DICOM image produced by the scanner). The oblong voxel is rotated, but still in the correct hemisphere (in other words, I cannot recover the true orientation via applying np.rot90 to the image voxel array)

Is there something I'm missing? Unfortunately I cannot share the data as it is from a clinical trial.

Best,

Rich