Image Registration feature request

[sm-ak-r33]

• TIA Toolbox version: pip installed latest one
• Python version: 3.9
• Operating System: Windows 10

Description

I want to get the transformation matrix out of the image registration process so that after the registration is done, if I multiply the moving image with the matrix, I get the registered image without involving the fixed image in layout.

What I Did

I tried generating the matrix but failed, is it even possible?

[Abdol]

Hi @sm-ak-r33. If I understand your question correctly, you can use the DFBRegister.register() method, which returns the transformation matrix as a numpy array. Here is an illustrative snippet:

from tiatoolbox.tools.registration.wsi_registration import DFBRegister
df = DFBRegister()
dfbr_transform = df.register(dfbr_fixed_image, dfbr_moving_image, fixed_mask, moving_mask)

You can also refer to our WSI registration notebook for a full example.

[sm-ak-r33]

Thanks so much for your response. Actually, I was trying to infer, if note a specific nuclei or point in the fixed image and multiply that point's co-ordinates with the transformation matrix would I get the same nuclei in the moving image or not. I will let you know if that works.

[Abdol]

@sm-ak-r33 let us know if you still need help with the above by re-opening this issue. Closing for now...