Mrtransform: [WARNING] the input linear transform is not orthonormal and therefore applying this without the -template option will mean the output header transform will also be not orthonormal

1

Hi all,

I am registering a T1w image to a B=0 image using the following code:

fast ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_brain.nii.gz

flirt -in ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_mean_b0_brain.nii.gz \
	-ref ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_brain.nii.gz \
	-dof 6 \
	-omat ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_diff2struct_fsl_initial.mat

flirt -in ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_mean_b0_brain.nii.gz \
	-ref ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_brain.nii.gz \
	-cost bbr \
	-wmseg ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_brain_pve_2.nii.gz \
	-init ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_diff2struct_fsl_initial.mat \
	-omat sub-${SUBJECT}_run-01_diff2struct_fsl_bbr.mat \
	-schedule $FSLDIR/etc/flirtsch/bbr.sch
	
transformconvert sub-${SUBJECT}_run-01_diff2struct_fsl_bbr.mat \
	${OUTPUT_DIR}/sub-${SUBJECT}_run-01_mean_b0_brain.nii.gz \
	${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_brain.nii.gz \
	flirt_import sub-${SUBJECT}_run-01_diff2struct_mrtrix_bbr.txt
	
mrtransform ${T1_DIR}/sub-${SUBJECT}_run-01_T1w.nii.gz \
	-linear ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_diff2struct_mrtrix_bbr.txt \
	-inverse ${OUTPUT_DIR}/sub-${SUBJECT}_run-01_T1w_coreg_bbr.mif

I am getting this warning:
the input linear transform is not orthonormal and therefore applying this without the -template option will mean the output header transform will also be not orthonormal

The results look good. The point of the registration is for ACT, so I am not massively keen on regridding to the B=0 grid.

This is the mrinfo output:
image

Can I safely ignore the warning? If not, any suggestions?

Claude

2

The second flirt call uses the default dof of 12 which results in a non-rigid transformation. We recommend rigid-body only registration. (You can assess the non-rigidness using transformcalc decompose sub-${SUBJECT}_run-01_diff2struct_mrtrix_bbr.txt).

Applying a non-orthonormal transformation to the image header can cause trouble for NIfTI files but is OK with .mif files if the information is interpreted in scanner space which as far as I know ACT does (@rsmith correct me if this is wrong).

1 Like