Hello, AFNI experts,
May I ask for advice on normalizing the result of 3dDeconvolve?
I am trying to normalize GLM results spatially on the MNI template. However, it seems that the results are not on the MNI template correctly.
I utilized this codes.
-prefix $rootDir/$SBJ/glm_normalized/$SBJ'_func_dstscvr_normsm4_PPI_Trials_bucket_fef_'$HEMI'_deobl+orig' $rootDir/$SBJ/$SBJ'_func_dstscvr_normsm4_PPI_Trials_bucket_fef_'$HEMI'+orig'
-dpar $rootDir/$SBJ/glm_normalized/$SBJ'_func_dstscvr_normsm4_PPI_Trials_bucket_fef_'$HEMI'+orig' \
-dxyz 1 \
Because the AFNI suggested to deoblique the dataset, it was performed before conducting the normalization. All the options of resampling methods (NN, Li, Cu, Bk). Unfortunately, all the resampling methods were not helpful. Furthermore, -dxyz value was also changed to 2 or 3. It was not helpful also. In this situation, what can I do to resolve the issue?
Hi, Shinyoung Jung-
Typically, one wouldn’t directly align a statistical dataset to a template, because they statistics map doesn’t have clear anatomical features (the kinds of data contained there are effect estimate and statistics blobs that might extend over structural boundaries).
For FMRI, during processing prior to statistical mapping, we might align the subject EPI to their anatomical dataset (creating the EPI-anatomical transformation), and then that anatomical dataset to the MNI template (creating the anatomical-template transformation), and combine/concatenate those transformations procedures into a single one (EPI-to-template transformation) to apply to the EPI so that it overlays well on the template. Then, when GLM modeling is performed, the results are already well aligned to the template, having used structural information in the EPI and anatomical datasets.
This is discussed, along with other alignment procedures that are part of standard processing and also concatenated into that single EPI-to-template transform, here:
And many other education videos are also on that AFNI Academy channel:
And actually, it is generally much more convenient to combine the alignment, statistical modeling, and other FMRI processing steps into a single pipeline that can be adjusted with options to do exactly what you want. In AFNI, this would be done with afni_proc.py, which is described here:
… and there are several starter examples in the afni_proc.py help, too: