Computer Vision News - August 2020

support available in the field for start- ups or spin-out companies to manage patient data coming out of hospitals , such as how it works, how to transfer it, how to license it. Even so, he hopes that in five years we may start to see products on the market with these kinds of AI-created biomarkers . Thinking about next steps, David finds interpretability a particularly interesting area: “Now that we have a system that works, we want to go back to the image and find out where the visual evidence is that is guiding the prediction. If we can show doctors what the algorithm is paying attention to, maybe they can figure out what is going on. This is using AI to do science . Until now, it was only doctors looking for these patterns; now we can use AI to accelerate the process.” Explainable AI is a hot topic nowadays, with entire research departments dedicated to working on it, and it is of particular interest to those working in computational pathology. He thinks it will be essential in the medical domain from the perspective of certification and informing the public. Runner-up for Best Paper Award 22 Best of MIDL 2020 “We spend millions on developing these complicated algorithms, but it’s important that we use them to improve patient care. That’s why I do it. To help people.”