N4BiasFieldCorrection for T2w?

feilong · March 7, 2024, 6:31pm

Hi fMRIPrep folks,

I wonder if the T2w image in fMRIPrep’s output FreeSurfer folder has been corrected using N4BiasFieldCorrection.

My understanding of the code is that it was performed for the T1w images (see code here, but I didn’t find similar code for the T2w image, so I guess the answer is no?

For some background information, I’m trying to test the Synth method for fieldmap-less SDC, and it requires corrected T1w and T2w as input.

Thanks,
Feilong

effigies · March 7, 2024, 7:30pm

Hi Feilong. Yes, T2w images are N4 corrected. They’re merged in the same way as T1w here:

github.com

nipreps/smriprep/blob/2355035d18d886cef5b6f0f4771e00c607d897a1/smriprep/workflows/anatomical.py#L1124-L1130


      
          t2w_template_wf = init_anat_template_wf(
              longitudinal=longitudinal,
              omp_nthreads=omp_nthreads,
              num_files=len(t2w),
              contrast='T2w',
              name='t2w_template_wf',
          )

The comments in init_anat_template_wf are still T1w-specific, but the same workflow is used for both modalities. If we start correcting PDw images, we’ll probably do the same there.

feilong · March 7, 2024, 7:42pm

That makes a lot of sense. Thanks for the clarification!