[ITK-users] SimpleITK landmark based initialisation (and image visualisation)

[Sami Koho]

Hi,

I was thinking of adding a manual landmark based initialisation to my image registration method, as the automatic initialiser does not reliably work in my current application. I’m registering correlative microscopy images with very large shift and rotation between moving and fixed images — usually the moving image also covers only a small part of the area shown in the fixed image. By specifying the initial offset and rotation manually, registration works nicely, but this is a bit cumbersome, because I have to figure out the approximate offsets first.

The problem is that I do not seem to have the sitk.LandmarkBasedTransformInitializer() available in my SimpleITK version. It was used in the notebook http://insightsoftwareconsortium.github.io/SimpleITK-Notebooks/67_Registration_Semiautomatic_Homework.html. I have SimpleITK 0.9.1 (Compiled Sep 28 2015) that I have installed on Anaconda according to the Wiki instructions. I am using Python 2.7. Would I possibly need some other version of SimpleITK? How could I get it? 

Alternatively, is there some other simple method for calculating a spatial transformation from control point pairs in SimpleITK? I can of course do that in Numpy, but I wouldn’t like to overly complicate things.

Another question, in the notebook above you are using a visualisation module named “gui" to show nice interactive plots, from which it is possible to make landmark selections as well. Is that by chance publicly available somewhere?  

Best,

Sami


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/insight-users/attachments/20160712/91c46c02/attachment.html>