I am looking for a loss function sensitive to edges for medical image quality enhuncement

Hello,
For an image-to-image translation task in which I want to generate high-quality images from low-quality images ( MRI Images), I need a loss function to highlight the edges and generate images with sharper edges.
Do you have any recommendation for selecting the desired loss function between Pytorch’s loss function??
https://pytorch.org/docs/stable/nn.html#loss-functions
I really appreciate it if anyone can even provide me the code of any predefined loss function for this task.

Thanks

If you want to make low-quality images into high-quality images and detect the edge of it, I’d recommend the image super-resolution models(such as FSRCNN or SRGAN) and then apply edge-detecting filters like Sobel filter.

Also, you can manually detect the edges from any direction by using image gradient

1 Like