Hello,

I have been trying to make sense of gantry tilted CT DICOMs and I might have come to a dead end.

Some background:
My understanding is that axial volumes are affected by the tilt as they are primarily acquired. The tilt calculated using the direction cosine matrix or the DICOM tag "Image Orientation", and the one mentioned under the tag "Gantry/Detector Tilt" do not always match, and I have been using the former when this is the case. The corrective transformation is a shear in the z-direction (tangent of the tilt), and my code seems to work (at least visually).

Questions:
1) When the sagittal and coronal volumes are reconstructed, this correction seems to be done automatically (in some cases) because the direction cosine matrices on the PACS do not indicate any tilt or shear, even though the axial series from the same study indicates a tilt. Is this correct?

2) In some cases, the axial series indicates a tilt that is completely different from the tilt computed using sagittal and coronal series from the same study. My understanding is that <Zi,Zp> in the axial series = <Yi,Zp> in the coronal series = <Yi,Zp> in the sagittal series (i indicates the image coordinates and p indicates the patient coordinates ; I have not taken sign into consideration here). But this is not the case in some of the studies I am working with.

Can someone please explain? There seems to be no documentation on this sort of information anywhere.

Thank you,
Sharada