Hi AFNI world,
I want to normalize a partial volume to the MNI using the code below:
@auto_tlrc -base MNI_caez_N27+tlrc -input 5004_UNI+orig
align_epi_anat.py -anat 5004_UNI+orig -epi 5004_npu1_nat+orig
-epi_base 6 -child_epi 5004_npu2_nat+orig.HEAD 5004_npu3_nat+orig.HEAD
-ex_mode quiet -epi2anat -suffix _altest
5004_UNI+orig corresponds my structural image, 5004_npu1_nat+orig corresponds to the first run of my fMRI task, 5004_npu2_nat+orig.HEAD and 5004_npu3_nat+orig.HEAD correspond to run2 and run3 of the task in their native space.
The script is running smoothly, however, when I observe the output image (see screenshot attached) I don’t understand why I observe these yellow signals a little bit outside of the brain. Am I doing something wrong?
Any help would be appreciated.
I believe that the yellow bits outside the brain are just a consequence of the programs using ‘wsinc5’ as the interpolating kernel when regridding the data (i.e., applying the warps). That kernel has plusses and minuses when compared to, say, a cubic spline: on the plus side, it preserves more contrast/sharpness within the volume; on the minus side, it will introduce tiny ringing, which will typically be seen just outside the brain (because the ringing is very small/unnoticeable when compared with the inner image). In general, for warping most non-integer-valued volumes, using this wsinc5 kernel is probably still the best way to go—in practice, that minimal ringing won’t really affect any calculations. It will just look weird.
I wonder if you might be able to use afni_proc.py to do this calculation, so you don’t have to manage the alignment pieces? I suspect that should still be possible and easier, at the end of the day.
And also, for most applications we recommend using full, nonlinear alignment between a subject anatomical and a template volume (@auto_tlrc is essentially just an affine alignment). That should give much better overall alignment, helping voxelwise group statistics as well as reference point creation or atlas registration for individual subjects.