Print

Print


Hi - registering binary images is strongly discouraged - you should register the full images or brain-extracted originals, and then apply those transforms to binary masks if desired.

Cheers.


On 19 Nov 2012, at 03:18, Longchuan Li wrote:

Hi, FSL experts

I am trying to register two binary images of the brain with slightly different outlines using FLIRT and for some reason, I could not succeed. The "-in" image is a white matter binary image in FreeSurfer space with 256*256*256 and resolution of 1mm isotropic. The reference image is in FSL space with very similar outlines (filled.nii.gz generated using FreeSurfer and registered to subject's FSL T1 space), but is still slightly different. It has the dimension of 220*220*160 with 0.5mm isotropic. It should be noted that these two images have identical resolution and the resolution of the input image was changed to 1mm as required by FreeSurfer. So, my questions are:

(1) what options are optimal to register two binary images  with "different" resolutions and orientations in FLIRT?
I used the following:
flirt -in white.nii.gz -ref filled2str.nii.gz -out white2str -searchrx -180 180 -searchry -180 180 -searchrz -180 180 -dof 7

(2) If I want to change the resolution of an image from 0.5mm isotropic to 1mm isotropic, what are the correct way to do it? Do I need to change both pixdim1,2,3 and the diagnoal scaling of the transformation matrix in the header information?

Many thanks in advance!

Longchuan


---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------