Hi, Sheerinsitara: I think that visual question answering on medical images is a very good (and important) topic for research and in practice. However, I think that there are two major problems here. First, finding or creating such datasets for study seems to be a difficult task. Second, I think that doctors need a model which only only gives the answer but also has to provide explaination on its reasoning or prediction. This is so called Explainable AI that is also a tough research area.