I’ve recently started playing some open-source software (STIR) to reconstruction SPECT projection data. While trying to recast the images into DICOM format, I’ve run into an issue regarding how to determine the Image Position Patient, Image Orientation Patient, and Spacing Between Slices attributes of the DICOM file.

I am trying to determine which bits of information in the DICOM header of SPECT projection data can be used to determine the attributes mentioned above, but I am failing to explain the values that I am seeing for various commercial SPECT(/CT) systems.

I expected (or perhaps, more accurately, hoped) that the Position Patient attribute (e.g. FFS, HFS, etc.) would be enough to indicate the Image Orientation Patient attribute, but that is not the case. Similarly, I thought that there would be a relationship between the sign of the Spacing Between Slices attributes and whether or not the patient was positioned head- or feet-first in the scanner – but I’ve found some studies that seem to violate that as well. Similarly, I have been unable to find a relationship between the Image Position Patient attribute of the projection data with that of the reconstructed image data.

I suspect that either different commercial systems are treating these data in non-standard ways and/or the relationships that I am seeking are simply less intuitive that I had assumed (hoped).

As I approach my wits’ end, I thought I’d reach out into the ether for help and see if something comes back.

Thanks in advance, ethereal OS user.