Computer Vision News - June 2018

Finally, can you tell us about your regular work besides the challenge? Medical segmentation is one of the main topics of my research. I use MRI images of the prostate to segment the prostate into the two main areas – the transitional zone and peripheral zone. I’m more a Machine Learning person, that is my background, but now I’m trying to apply this very theoretical background acquired during my previous experience at the University of Pisa in the field of medical imaging. What I’m trying to do is to build a computer-aided system which can be used to better classify the lesion inside the prostate and avoid unnecessary prostate biopsy. The idea is to build a system that can first identify where the prostate is, then identify if there are some pathologies inside this area. I do this by applying different techniques, from deep learning to more traditional classification systems. It is a tough area because the images are very noisy and there is a lot of variation between patients. It is not easy even for the human eye to spot differences between different types of lesions inside the prostate. 24 Challenge: Medical Decathlon Challenge Computer Vision News “…to build a computer- aided system which can be used to better classify the lesion inside the prostate and avoid unnecessary prostate biopsy!”