Intensity inhomogeneity correction

AFNI Message Board
1

Dear experts,

I would like to correct the inhomogeneities caused by spatial intensity variations in EPI volumes (as performs the “fast” function in FSL).
Is there an equivalent function in AFNI to do this?

Thank you very much in advance.

Marina.

2

3dUnifize is the program we use for intensity inhomogeneity correction; however, we typically use this on the anatomical datasets and not on the EPI data. You may use this on EPI data for an alignment target, but not for GLM analysis. The GLM analysis pipeline with afni_proc.py will typically include a scaling step, where each voxel is treated independently of the others. No grand mean scaling is performed only voxelwise scaling, so that’s less of a concern in AFNI.

3

Thank you for your reply Daniel.

I should clarify that the EPI volumes don’t have severe or evident artefacts and the reason why we would like to correct the inhomogeneities is to improve us much is possible the alignment of the EPI volumes. Considering these facts:

  1. Do you recommend to apply 3dUnifize only in the case that the inhomogeneities are evidents?
  2. In case that you think we should always apply 3dUnifize in our preprocessing (even if the inhomogeneities are not evident), when should we apply it? Before EPI volume alignment or before the alignment of de EPI volumes to the anatomical?

Best,
Marina

4

We almost never do this kind of correction on EPI datasets, and then only for alignment, not for the GLM analysis. Unless there are very strong shading or coil artifacts, I would avoid doing this. If you do want to use it in our standard pipeline, afni_proc.py does include an option to unifize the EPI data only for alignment with “-align_unifize_epi yes”.

5

Doubts solved! Thank you so much for the recommendations Daniel.