FLIRT registration of FLAIR into MNI space results cropped and off-center

Benjamin_Leger · March 13, 2024, 2:15pm

Hello everyone,

I am trying to register a brain-extracted T2 FLAIR image into MNI space using FLIRT.
The output image is totally distorded and cropped of and i don’t understand why.

The fslinfo of my file :

filename	FLAIR_2_Brain.nii.gz
sizeof_hdr	348
data_type	FLOAT32
dim0		3
dim1		380
dim2		480
dim3		340
dim4		1
dim5		1
dim6		1
dim7		1
vox_units	mm
time_units	s
datatype	16
nbyper		4
bitpix		32
pixdim0		1.000000
pixdim1		0.500000
pixdim2		0.500000
pixdim3		0.500000
pixdim4		8.000000
pixdim5		0.000000
pixdim6		0.000000
pixdim7		0.000000
vox_offset	352
cal_max		0.000000
cal_min		0.000000
scl_slope	1.000000
scl_inter	0.000000
phase_dim	0
freq_dim	0
slice_dim	0
slice_name	Unknown
slice_code	0
slice_start	0
slice_end	0
slice_duration	0.000000
toffset		0.000000
intent		Unknown
intent_code	0
intent_name	
intent_p1	0.000000
intent_p2	0.000000
intent_p3	0.000000
qform_name	Scanner Anat
qform_code	1
qto_xyz:1	0.499851 -0.000183 -0.012216 -103.352325 
qto_xyz:2	0.001787 0.495659 0.065720 -112.136017 
qto_xyz:3	0.012086 -0.065744 0.495511 -42.285835 
qto_xyz:4	0.000000 0.000000 0.000000 1.000000 
qform_xorient	Left-to-Right
qform_yorient	Posterior-to-Anterior
qform_zorient	Inferior-to-Superior
sform_name	Scanner Anat
sform_code	1
sto_xyz:1	0.499847 -0.000183 -0.012216 -103.352325 
sto_xyz:2	0.001787 0.495659 0.065720 -112.136017 
sto_xyz:3	0.012086 -0.065744 0.495511 -42.285835 
sto_xyz:4	0.000000 0.000000 0.000000 1.000000 
sform_xorient	Left-to-Right
sform_yorient	Posterior-to-Anterior
sform_zorient	Inferior-to-Superior
file_type	NIFTI-1+
file_code	1
descrip		2203.12-dirty 2024-02-01T16:17:47+00:00
aux_file

The template file :

filename	MNI152_T1_0.5mm.nii.gz

sizeof_hdr	348
data_type	UINT8
dim0		3
dim1		364
dim2		436
dim3		364
dim4		1
dim5		1
dim6		1
dim7		1
vox_units	mm
time_units	s
datatype	2
nbyper		1
bitpix		8
pixdim0		-1.000000
pixdim1		0.500000
pixdim2		0.500000
pixdim3		0.500000
pixdim4		1.000000
pixdim5		0.000000
pixdim6		0.000000
pixdim7		0.000000
vox_offset	352
cal_max		250.000000
cal_min		0.000000
scl_slope	1.000000
scl_inter	0.000000
phase_dim	0
freq_dim	0
slice_dim	0
slice_name	Unknown
slice_code	0
slice_start	0
slice_end	0
slice_duration	0.000000
toffset		0.000000
intent		Unknown
intent_code	0
intent_name	
intent_p1	0.000000
intent_p2	0.000000
intent_p3	0.000000
qform_name	MNI_152
qform_code	4
qto_xyz:1	-0.500000 0.000000 -0.000000 90.000000 
qto_xyz:2	0.000000 0.500000 -0.000000 -126.000000 
qto_xyz:3	0.000000 0.000000 0.500000 -72.000000 
qto_xyz:4	0.000000 0.000000 0.000000 1.000000 
qform_xorient	Right-to-Left
qform_yorient	Posterior-to-Anterior
qform_zorient	Inferior-to-Superior
sform_name	MNI_152
sform_code	4
sto_xyz:1	-0.500000 0.000000 0.000000 90.000000 
sto_xyz:2	0.000000 0.500000 0.000000 -126.000000 
sto_xyz:3	0.000000 0.000000 0.500000 -72.000000 
sto_xyz:4	0.000000 0.000000 0.000000 1.000000 
sform_xorient	Right-to-Left
sform_yorient	Posterior-to-Anterior
sform_zorient	Inferior-to-Superior
file_type	NIFTI-1+
file_code	1
descrip		FSL5.0
aux_file

I am using this flirt line, via the GUI :

flirt -in /Users/.../FLAIR_2_brain.nii.gz -ref /Users/.../MNI152_T1_0.5mm.nii.gz -out /Users/.../FLAIR_2_registred.nii.gz -omat /Users/.../FLAIR_2_registred.mat -bins 256 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 12  -interp trilinear

The result totally cropped and distorded and i can’t figure out what went wrong:

Could you help me understand the issue ?
Thank you so much !

Steven · March 13, 2024, 2:20pm

Hi @Benjamin_Leger and welcome to neurostars!

One issue is that you are directly registering a FLAIR image to a T1. What you should do, if possible, is to find a T1 in the same space as your FLAIR image, register that T1 to MNI, and then apply the same transformation to your FLAIR.

Can you also confirm that the MNI T1 is brain-extracted? Do the images have the same orientation (e.g., LPS, RAS) are both images labeled correctly when loaded into an image viewer?

Best,
Steven

Benjamin_Leger · March 13, 2024, 2:28pm

Hi Steven,

Thanks for the tip i will try to use a T1 first !
I am afraid images are not the same orientation, as my template is labelle “right-to-left” in sform et qform_xorient whereas my file is “Left-to-right” but i have no clue how to change this, i have already tried reorient2sd but doesn’t change anything

this is what the images (my MRI and the template) look like when put in a viewer

Steven · March 13, 2024, 2:30pm

Hi @Benjamin_Leger,

As long as the images are labeled correctly in an image viewer (e.g., what is labeled as left is really left in the brain) I would try running fslreorient2std on your images before registration, just to ensure orientations are the same.

Best,
Steven

Benjamin_Leger · March 13, 2024, 2:36pm

I maybe have a clue : my images are routine MRI exported from a viewer, and by radiological convention right is at left and left at right, whereas my template is anatomical (right is right and left is left). How could i reverse my MRI ?

Steven · March 13, 2024, 2:39pm

As long as the image labels are correct in a viewer, fslreorient2std should do the trick.

Best,
Steven

Benjamin_Leger · March 13, 2024, 4:52pm

A little bit better, by using a T1 and mutual information as cost function (correlation ratio was the same problem as FLAIR).
But the brain is too large compared to the mni template, with the brain taking the whole size of the template instead of only the “brain part” of the template. What do you think is the problem ?

Here an image of the preprocessed mask transparent over the mni template

Steven · March 13, 2024, 6:13pm

Are both images skull stripped? It looks like brain extraction did not go great for either the over or underlay.