[Insight-users] Liver registration in MRI slices

[Lopez Saratxaga, Cristina]

Dear all,

I've several sets of MRI images, each one with different sequences and multiple slices within each.  You can see attached an example of a slice (axial view).

Our goal is to produce an automatic segmentation of the liver (located at the left side) taking into consideration the information not only of the different slices of one sequence, but the information contained in various sequences in order to create a 3D model of the liver. One colleague is already working on in, but in order to make sure that the segmentation is done as best as possible, we need to perform a registration of the slices through the various sequences.

At first I tried with basic transformation as the Affine of Deformable, applying diverse metric and optimizer, but the segmentation didn't improve in any respect. These transformations work fine, but the problem is that the registration is mainly made in the contours of the image (2D registration performed better than 3D registration). What I would like, would be to just focus in the region of the slice where the liver should be located and focus the registration on the liver, however, I'm afraid that I don't know how to perform this.

I've read something about registrations based of features. The landmark based registration sounds interesting but it seems that it's necessary to define a set of reference points both if the fixed and the moving image, something that I don't have. At the most, I could have some approximate reference points for the fixed image, but not for the moving one. Additionally, I've also read something about masks and ROIs. I tried the example ImageRegistration12 with a binary mask, however, the results were not as expected (maybe I was not using well the mask).

At this time, I'm a bit lot with this problem and I really don't know how to proceed with the registration problem, so any idea on how to deal with this problem will be really appreciated!
Many thanks in advance for your Help!!

Kind regards,

Cristina

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20130729/e18fb912/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: MR000245.png
Type: image/png
Size: 43793 bytes
Desc: MR000245.png
URL: <http://www.itk.org/pipermail/insight-users/attachments/20130729/e18fb912/attachment.png>