I have multiple sMRI data sets where each subject can have multiple sessions and multiple runs within these sessions. I would like to use MRIQC to get an overview of the images and eventually run MRIQC’s classifier on all data sets to get the ‘best image’ for each subject. I have a few general questions that I couldn’t solve myself by just browsing MRIQCs documentation.
1.) It seems that MRIQC and fMRIPrep are somehow related, that is, just like MRIQC, fMRIPrep also outputs quality measures and visual reports for my s/fMRI images. It even looks like fMRIPrep uses MRIQC in the background (or at least they just share some nipype workflows)? Or maybe I got on the wrong path here and they just share the same visualization frontend which makes them ‘look similar’? In this this neurostars post, @oesteban recommended running MRIQC first and then running fMRIPrep only on the images with ‘good quality’. What is the rationale for following this order when fMRIPrep also gives me visual reports? Or in other words, where do MRIQC and fMRIPrep have overlaps with regards to quality control and where do they differ?
2.) When looking at the overview of the anatomical workflow from MRQIC it seems to follow a canonical sequence of sMRI preprocessing steps (skull-stripping, normalization, segmentation). However, MRIQC does not output a preprocessed image, although based on the documentation it looks like it theoretically could? MRIQC also seems to be ‘alarmingly’ fast (too fast than one would expect for a pre-processing pipeline), so my intuition is that this preprocessing pipeline is kind of made for only serving the purpose of outputting QC measures but not providing a preprocessing pipeline in the strict sense? Is that right?
TLDR; What are the similarities and differences between MRIQC and FMRIprep?