So for instance, I’ve been sent intra-operative dMRI data with the patient acquired in the ‘decubitus left’ position (on their side). However, the dedicated tags in the new DICOM standard ((0018,9087) Diffusion b-value, and (0018,9089) Diffusion Gradient Orientation) also expect them to be provided relative to the PCS – at least, that’s what the standard suggests (could be clearer though: “The direction cosines of the diffusion gradient vector with respect to the patient”). ![]() GE stored them relative to phase-read-slice, so that doesn’t apply. I’m pretty confident this holds for Siemens, not sure about Philips (not seen any indication to the contrary yet, but I’m not sure I’ve come across non-HFS data yet). The PCS is the frame that any position or vector information in the DICOM headers should be provided relative to – including the DW gradient orientations. The idea is to ensure that the anatomical orientation labels (LR, AP, IS) retain their expected meaning relative to the patient independently of how they were positioned in the scanner. I would expect this biped/quadruped entry to affect the PCS, same as putting the patient in feet-first, etc. ![]() In DICOM, all coordinates are provided relative to the DICOM patient-centred coordinate system (PCS), which importantly is not fixed relative to the scanner axes. I think it’s a matter of interpretation, particularly of what the NIfTI standard frame actually means. ![]() However, I suspect this would require an orthogonal rotation such as the one suggested if the data is from Siemens, Philips or UIH.
0 Comments
Leave a Reply. |