Magnitude image(s) in MRI phasediff fmap

[Lestropie]

Seeking clarification on "Types of fieldmaps" - "Case 1: Phase-difference map and at least one magnitude image". I only have conceptual knowledge of this technique, not hands-on experience, so want to make sure my understanding is correct before proposing a change.

Based on the current phrasing:

where the REQUIRED _phasediff image corresponds to the phase-drift map between echo times, the REQUIRED _magnitude1 image corresponds to the shorter echo time, and the OPTIONAL _magnitude2 image to the longer echo time.

, it would be reasonable to infer that, if it is only possible to provide a single magnitude image, that image MUST correspond to the shorter of the two echo times.

I'm looking at retrospective Siemens data where the sequence has emitted two DICOM series. The second looks to be a phase difference, hence my flagging Case 1 as applicable. However it seems that the single magnitude image is some convoluted interleaved stacking of the first and second echoes. dcm2niix has to have the "merge 2D slices" option utilised to produce something that looks like a whole head. Therefore, in this scenario, there is only a single magnitude image volume available; but it does not correspond exclusively to the first volume.

If this is a regular manifestation of such data, I will propose a modification to the specification. Essentially, if there are two magnitude images, then _magnitude1 should indeed be the earlier of the two echoes; but if only one magnitude image is available, then that should be stored as _magnitude1 regardless of the constitution of such.

Whether this is appropriate is predicated on how such data are used. I believe that the corresponding magnitude image is only used as a registration target for any images to which the estimated inhomogeneity field is to be applied; in which case, whether that registration target is the first echo, the second echo, or some weird interleave of the two, should not be consequential.

Am I missing something fundamental here?

[effigies]

This matches my understanding. The magnitude images are only used as registration targets, and if you get one, it's magnitude1. If you get two, the shorter echo should be magnitude1.