Hi Afni Experts:
I ran into an odd issue using code that I have used before. I used to use 3dcalc to combine GM/WM/CSF brain masks and subtract the saggital sinus mask out I created. I would then use the 3dmask_tool command to fill in any holes in the resulting file:
3dcalc -a mask_c123456.nii -b mask_c5.nii -expr a-b -prefix WB.nii
3dmask_tool -input WB.nii -dilate_input 2 -2 -fill_holes -prefix mask_WB.nii
I would import this mask into freesurfer and view using the following and everything overlayed perfectly with the underlay files:
mri_convert -i mask_WB.nii --in_type nii -o mask_WB.mgz --out_type mgz
mri_vol2vol --mov mask_WB.mgz --targ brainmask_FS.mgz --regheader --o mask_WB_FS.mgz --no-save-reg --nearest
tkmedit to view the images as I use this mask in the freesurfer pipeline.
Study 1:
When viewing in freesurfer everything looked perfectly aligned and my scan acquisiton for this particular study was taken without any oblique angles.
Used the following:
Ubuntu 10.04LTS 64 bit
Precompiled binary linux_openmp_64: Jan 31 2014 (Version AFNI_2011_12_21_1014)
Study 2:
I am taking images with an oblique angle for this particular study.
This system with issues has the following:
Ubuntu 14.04LTS
Precompiled binary linux_openmp_64: Aug 23 2014 (Version AFNI_2011_12_21_1014)
When taking the WB.nii image to freesurfer using the steps above (omitting 3dmask_tool) everything again looks perfectly aligned. However when I use 3dmask_tool to fill in holes on the mask derived from oblique scans, this is where I run into issues. The images are always shifted and not aligned with the T1 imported image after converting to freesurfer. When running 3dmask_tool with this same input (WB.nii) on the other system (Ubuntu 10.04LTS/binary Jan 31 2014 from the study above) the image looks correct in freesurfer and overlays perfectly with the T1. Additionally, when using Precompiled binary linux_openmp_64: Jan 31 2014 (Version AFNI_2011_12_21_1014) as opposed to Precompiled binary linux_openmp_64: Aug 23 2014 (Version AFNI_2011_12_21_1014) on the Ubuntu 14.04LTS system, everything looked correct as well.
There seems to be an issue or change in 3dmask_tool from precompiled Jan 31 2014 version to August 23, 2014 version which seems to be causing this error but I thought it was worth mentioning as I do not know if this change is localized to the openmp version or affecting all packages. Hope the description helps for debugging and please let me know if there is a fix I can implement or if it may be taken care of in the next release.
Thanks,
Ajay
