I am preprocessing NAPLS3 dataset, which has older gradient echo fieldmaps from GE scanners.
The resulting phase images have bands in them that may be related to improper unwrapping. This is the case for both the phase images output from dcm2niix and for phase images generated by using the real and imaginary images output from dcm2niix (fslcomplex -complex followed by fslcomplex -realphase). I aim to generate a phasediff image for downstream qunex starting with hcp_prefreesurfer.
Examples of the phase images are attached below from one subject in site 4. Any suggestions on how to fix this data and ultimately generate unwrapped phasediff images that are suitable for qunex hcp_prefreesurfer?
I talked with the person who has the most experience with GE fieldmaps on our end. He says that it looks like ph images have some kind of an artefact, maybe due to subject motion or some issues in the acquisition sequence. _phase images look quite OK, besides the wide bands.
GE fieldmaps apparently come in many formats and not a single one, so getting pipelines to support all that is tricky. He said that you seem to be on the right path. He suggests that you try preparing the fieldmaps as needed by the pipelines by using your _phase images and then run the HCP minimal preprocessing on a test sessions and see where that gets you. If things work well should be visible by overlaying relevant images and checking the distortion correction.
