I just recently finished going through all the lessons and videos in Part 2. Throughout the course, I have been thinking of ways that I could use the different frameworks to tackle problems that I see. One of these problems that I would like to work on is the suppression of bone shadows on chest x-rays, but I’m unsure of the best architecture to train on. I feel like style transfer and super resolution both have parts that could be useful, but I’m really unsure how to approach this. Any thoughts on where to start?
Here’s an example of and image with this technique implemented: