To extract texture features from specific ROI using PyRadiomic, I tried two different methods:
Use fMRIPrep with --anat-only option
Simply run a Nipype workflow with RegistrationSynQuick interface
This is the distribution of one features across all subjects:
I’m wondering what might be the reason why images processed by fMRIPrep have a different distribution with smaller variance. Should I use sMRIPrep instead?
Here are two images from the same subject with different methods: