[Insight-users] A question about jacobian modulation.

[johnjtchen at aol.co.uk]

 Dear All,

When I was trying to apply Jacobian modulation on nonrigid registered 
(SymmetricforceDemon) repeat scan, I found that the modulated image 
(voxel intensity * det(Jacobian)) does not look realistic but the 
registration result is almost perfect from image subtraction. The 
nonrigid registration and Jacobian modulate was done in both binary 
(segmented white + grey matter) and in original (full range intesity, 
skull stripped). In original image, there are some high intensity voxels 
around white matter region so the modulated image looks awkard. The 
intensity range of deteminant of Jacobian with the original is [-6.72, 
3.743] and with segemented image is [-0.49, 2.899]. The smooth options 
in SymmtricDemon were all turn on and tried various StandardDeviation 
settings. The Strange thing is that the deformations within white matter 
region is really noisy (I have appplied Histogrammatch and bias 
correction before NR starts), which is not I am expecting in the 
homogeneous region.
And then I took divergence and norm image from deformation field, I 
found that the divergence and norm over the white matter region are much 
greater then venticle region (I am trying to measure longitudinal 
difference).

I've also tried LevelSet and Bspline registration. THe deformation field 
and registraion results are almost as good as baseline scan like 
Demons.The defromation fields are totally different from each other. In 
T. Rohlfing's paper [1]. Although the det(Jac) are diffrent from various 
methods but the Jacobian integration does correpond the volume 
difference. But I just cannot get similar result!

I really cannot figure out why, so can someone give me some advice?

Thank you very much.

*[1] T. Rohlfing*, “Transformation model and constraints cause bias in 
statistics on deformation fields,” in /Medical Image Computing and 
Computer-Assisted Intervention - MICCAI 2006: 9th International 
Conference, Copenhagen, Denmark, October 1-5, 2006, Proceedings/, R. 
Larsen, M. Nielsen, and J. Sporring, Eds., Berlin/Heidelberg, 2006, vol. 
4190 of /Lecture Notes in Computer Science/, pp. 207-214, Springer-Verlag.




Best
John
________________________________________________________________________
Check out the new free AOL Email -- 2GB of storage and industry-leading spam and email virus protection.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20070110/025c1eb6/attachment.htm