Huge success for a team of researchers from the Biomedical Image Analysis group at the Department of Computing. The team of 11 PhD students, post-docs and academics ranked top on the prestigious, international computational challenge on brain tumour segmentation (BraTS). The BraTS challenge did run for the sixth time and this year more than 50 teams participated from a wide range of top universities and research institutes.
The BioMedIA team developed an approach called EMMA, which stands for Ensembles of Multiple Models and Architectures, using cutting-edge deep learning technology for image analysis. The goal of the challenge was to automatically detect and segment brain tumours in MRI scans, and EMMA obtained top ranking performance with unprecedented accuracy. The team is currently preparing a scientific publication describing their approach in detail and the implementations will be made publicly available as open source.
The full BioMedIA team includes Kostas Kamnitsas, Wenjia Bai, Enzo Ferrante, Steven McDonagh, Matt Sinclair, Nick Pawlowski, Martin Rajchl, Matt Lee, Bernhard Kainz, Daniel Rueckert and Ben Glocker.