users > Rigid registration weird test result
Showing 1-5 of 5 posts
Results per page:
Aug 8, 2018  08:08 PM | Zhilei Zhao - Princeton University
Rigid registration weird test result
Dear CMTK users,

My name is Zhilei. I'm trying to register some confocal images of antennal lobe. Since I just started to learn CMTK, the first thing I did is to run a test registration: put one confocal stacks in the refbrain folder, then use the Fiji to translate 20 pixels in X and Y, then register this transformed stacks with the following command:
registrationx -v --dofs 3 -o Registration/affine/OrcoA11_OrcoAfake_01_9dof.list --write-reformatted temp.nrrd refbrain/OrcoA11_01.nrrd images/OrcoAfake_01.nrrd

If everything works, I should get transformation only in X and Y. However, the transformation result seems unexpected:
Resulting transformation parameters:
#0: 7.400000
#1: 0.300000
#2: 0.400000
#3: 0.000000
#4: 0.000000
#5: 0.000000
#6: 1.000000
#7: 1.000000
#8: 1.000000
#9: 0.000000
#10: 0.000000
#11: 0.000000
#12: 53.305594
#13: 53.305594
#14: 34.997497

When I check the reformatted image, it seems rotated 180 degree. I attached a screenshot showing a stack of the ref image, float image and reformatted image at the same z depth. 

I feel I may get some fundamental things wrong. Your help is very appreciated!
Attachment: CMTK_test.png
Aug 9, 2018  10:08 AM | Greg Jefferis
RE: Rigid registration weird test result
Dear Zhilei,

I see no sign in the output parameters that it was trying to rotate the images. It seems to me that it did a 7.4 µm X displacement and a small Y and Z displacement – now of course that may not be so sensible, but that is what it says. I wonder if your images have any anatomical orientation information in the header, which is resulting in CMTK reorienting the images. Would you like to share the header information from your two input and one output images e.g. what you get by doing Fiji Image ... Show Info?

Best wishes,

Aug 9, 2018  12:08 PM | Zhilei Zhao - Princeton University
RE: Rigid registration weird test result
Hi Greg,

Thank you so much for help!

I attached the header info for refbrain, float image and reformatted stacks. The raw confocal stacks are in the tiff format, then I used Fiji to convert to nrrd format. Not sure this process may cause weird orientation change. 

Aug 9, 2018  03:08 PM | Greg Jefferis
RE: Rigid registration weird test result
Your refbrain has a line:

space: left-posterior-superior

This is an anatomical space. I am not sure how it got there. It looks like this image stack was not saved by Fiji. This seems to have somewhat confused cmtk since the other image has no anatomical space.

I recommend resaving this image from Fiji and then trying again.


Aug 9, 2018  08:08 PM | Zhilei Zhao - Princeton University
RE: Rigid registration weird test result
Yup! I just tried removing this space line in my refbrain and now CMTK gives transformation results that match exactly to my arbitrary translation. And the reformatted stacks also look normal now. 

Thank you so much,