In the ideal / simplest case, yes. However there are a lot of complications for what may naively seem like a trivial thing. Over and above potential issues with DICOM export / import, there’s non-standard patient positioning (e.g. scanner is configured as “supine” but the patient is actually on their side); non-trivial axis strides complicating the relationship between axis order in the file on disk and its interpretation; and in many cases the way that the axes are represented in file (even accounting for non-trivial strides) and the way the image is interpreted by MRtrix3 differs as it attempts to satisfy your statement above. And that’s not to mention that with oblique acquisitions, especially strongly oblique, image axes can never perfectly co-align with scanner spatial axes / anatomical axes, they can only correspond as best as is possible. It’s all a mess really.
See below the image from the original (left) and the reordered (right). I overlaid the numbers of what I thought were the axes in the input as presented by
mrview:
mrview’s ortho view mode isn’t intended to present projections in an order that’s somehow commensurate with axis ordering. Even if it did I’m not sure it would make sense even after becoming aware of such, since for each projection the “corresponding axis” is the one going in-out of the screen, i.e. the axis you can’t see in that particular projection, I don’t actually recall what the justification for that particular ordering is (if there is one).
Funny you mention that, because I did perform an X flip using
mrtransform -flip 0and compared both flies and I really couldn’t tell any substantial difference between the two. I felt like I was looking into the brain in a mirror, but still making sense at the axes levels. From your experience, what would you recommend in terms of an “independent” way to check left and right in the scan?
If it’s a concern for you and intrinsic markers are proving difficult, you could do a one-off scan with a physical fiducial. That will at least tell you which direction is left/right for any acquisitions that use an identical scanner configuration to that one-off. But you’d also want to validate all acquired data against that one-off to try to detect deviations from the acquisition configuration that may render it inapplicable.