Dicom conversion for images with gantry tilt

Hi Folks,

I have some data acquired on a Philips scanner which are distorted / stretched such that the z-axis is off centre. The below image was converted with dcm2nii(x?) via SPM.

Digging around in the dicom images on Horos gave this warning about gantry tilt.

This paper discusses gantry tilt but only in the context of CT scanners:


Does anyone have experience of correcting for gantry tilt in MR images? If so, does this look like gantry tilt to you, and do you have recommendations for tools to correct?

Many Thanks!


CT images are sometimes acquired with Gantry tilt, but MRI scans are always acquired with rectangular volumes. Your example is a MRI scan, so I do not think the original DICOM or initial NIfTI image had any skew. Many tools can only handle rectangular volumes, and can not handle images where the three axis are not orthogonal to each other (including all VTK based tools and many viewers). dcm2niix is not included with SPM. So the images might have been converted with SPM or converted with dcm2niix. Both are pretty robust conversion tools. My best guess is someone applied a spatial transform that applied this skew (e.g. old_norm) during a 12-dof affine transform where the source image was a poor match to the target reference template.

I would suggest running dcm2niix from the command line. If it detects Gantry tilt it ill generate a warning. It may also generate other warnings that will reveal core issues with these images. My guess is it will convert the images fine. You can check the spatial transformation matrix with fslhd or with MRIcroGL (open the NIfTI image, press the ‘Options’ button, select ‘Show Header’ look at the ‘Reorient’ tab).