XTRegister Same Channel Problem

[rosedever]

I am currently using the SimpleITK program for the main purpose of aligning serial cuts with the XTRegisterSameChannel with the SimpleITK Imaris Python Extension. I have used this without issue on multiple experiments but on my most recent iteration the alignment seems to be off.

To align the two images I completed the following steps:

1. Open the Imaris Extensions Driver
2. Set up Affline Registration of z-stack using common channel
3. Select the files and set the common prefix separator
4. Register the shared channel
5. Resample and save the combined image
6. Compute Correlation

These are the shared channel on the two cuts:

While the second image loses some signal towards the bottom of the image, the top I predicted to be able to align.

In the combined image, I have solely shown the same channel. However, it does not appear to have worked as the follicles shown on the top right of the image do not superimpose:

I think this is a a bug with an issue of aligning two images. Is there a problem with aligning two images that are not of equal size?

This occurs even while changing which image is the "Fixed Image" during same channel registration

I am using Python Version: 3.12.4

[zivy]

Hello @rosedever,

This is not a bug, just a failure of the registration to converge to the correct solution. Registration is not guaranteed to always yield the correct result.

With respect to your question about the image sizes, the sizes of the images do not have to be the same, but they are expected to be similar. Supposedly we are looking at the same object at the same magnification. If the overlap between the two images is minimal, we are not really looking at the same object, then the registration approach used by the XTRegisterSameChannel will fail. This doesn't seem to be the case here.

Please try to run the registration and double the value for "Samples per parameter" found under the "Advanced Registration Settings". If this doesn't work, possibly share your images and we can take a look.

As a side note, when displaying the combined result of registration it is better to use a different color for each image (in the example you provided both images are in green). To see why alignment is better visualized by using different colors see the image used in this post on the IBEX Imaging discussion.