Deep neural networks for image quality enhancement typically need large\nquantities of highly-curated training data comprising pairs of low-quality\nimages and their corresponding high-quality images. While high-quality image\nacquisition is typically expensive and time-consuming, medium-quality images\nare faster to acquire, at lower equipment costs, and available in larger\nquantities. Thus, we propose a novel generative adversarial network (GAN) that\ncan leverage training data at multiple levels of quality (e.g., high and medium\nquality) to improve performance while limiting costs of data curation. We apply\nour mixed-supervision GAN to (i) super-resolve histopathology images and (ii)\nenhance laparoscopy images by combining super-resolution and surgical smoke\nremoval. Results on large clinical and pre-clinical datasets show the benefits\nof our mixed-supervision GAN over the state of the art.\n