The question: Does QSIPrep use the method described in this paper?
I recently came across this paper (Wang et al., 2017) which seems to suggest that fieldmap-less distortion tended to outperform distortion correction based on an acquired fieldmap in terms of FA test-retest reliability. They describe the fieldmap-less method as
To evaluate the scenario where no field map was acquired (Figure 3) we used ANTs SyN nonlinear registration of the unweighted diffusion image to a bias corrected (using FSL FAST) and skull-stripped (using ROBEX, a robust, machine learning-based brain extraction system (Iglesias et al., 2011)) T1 image.
I know QSIPrep uses the same fieldmap-less distortion as fMRIPrep, but I cannot find a citation on fMRIPrep’s distortion correction webpage, although it looks like it also uses ANTs SyN. Can someone confirm whether or not QSIPrep/fMRIPrep use the method described in the paper?