Computer Vision News 24 Now, with an approach that works across different modalities and anatomies, the team modularized it so that representation learning with segmentation can be combined with any registration network. Testing it on two networks, a very basic CNN and a transformer architecture, showed that it generalized better in both cases. The paper’s success marks a significant step forward, but the researchers acknowledge that more work remains to be done to realize the approach in a real-world medical setting. They point to the need for further research into generalized solutions, such as nnUNet and TotalSegmentator for segmentation. MIUA is a modest but popular event on the conference calendar. Christoph and Mattias are keen to thank the organizers and tell us how much they enjoyed their time in Aberdeen and hope to return for future editions. Having scooped such a prestigious award at the event, has Christoph considered what made his paper a winner? “We can use advantages from one field to solve or find some kind of solution in another field!” Best Paper MIUA 2023
RkJQdWJsaXNoZXIy NTc3NzU=