Computer Vision News - August 2020

Victor and Francesco 15 Which one of the three is most likely to take home a Nobel Prize one day? Valentina is quick to answer. “The one you need to keep an eye on is Victor,” she tells us. “I’ll need to find something to wear first though – I never have anything to wear!” she laughs. After such a successful experience at MIDL 2020 , could the team take home the prize two years in a row? Valentina tells us this work is part of a bigger project. In fact, they had another paper at MIDL this year on task-based reconstruction for which Francesco was first author. “Sadly, we couldn’t win twice in one year!” she jokes. “That topic is showing big promise. It’s an optimization deep learning framework , so it allows us to optimize our reconstruction to be able to do some post-processing. What we do now is we create an image and this image is looked at by a radiologist, but what we can do is create an image that is automatically processed by a machine. If the pipeline is end to end, we can have an intelligent system that is able to recognize the abnormality on an accelerated image while you are in the scanner. We always talk about precision imaging and precision medicine , and that is exactly what this is.” By knowing the characteristics of the patient while they are in the scanner, you can change the protocol and the examination while they are on the table . If you know that the patient has a certain abnormality, you may want to add a particular sequence, or keep them there for a further examination. Currently, if patients need a further exam, they will be called back for another appointment once the radiologist has read the results. If that can all happen while the patient is on the table, it is a huge advantage. Valentina exclaims: “That is what is going to win Best Paper at MIDL next year!” Best of MIDL 2020