How can I extract sub image (pixels) using CUDA/C++?

This is my first time using CUDA C++, so I have idea but I am not sure if it’s correct.

I am working on high resolution imaging microscopy, so I want to extarct the SPF spots from image microscopy in parallel. The spots here are the sub regoins.

To do this, I want to transfer an image to device,an image is taken as a grid divided in (11x11) pixels blocks , that’s mean each block is sub regoins.

Any suggestions please .