Computational medical imaging and machine learning – methods, infrastructure and applications
Over the past few years there has been a dramatic development in areas associated to machine learning and artificial intelligence. This is caused by breakthroughs in “deep learning”, a collection of techniques that enable computers to uncover complicated patterns and connections in large data sets. Increased access to data (“big data”) and increased computational power has made it possible to use deep neural networks, which has become the state-of-the-art approach to a many key challenges in computer vision, language modelling and robotics. These developments have enormous potential also within medicine, where large data sets from health registers, images, biopsies and gene sequences are collected.
The goal of the project, Computational medical imaging and machine learning – methods, infrastructure and applications, is to develop, implement, disseminate and evaluate machine learning techniques in the analysis of medical images and image-related data.
To successfully incorporate machine learning in medicine, doctors and medical specialists have to take a leading role. Our project, initiated from the Department of Biomedicine at UiB and the Department of Computing, Mathematics and Physics at HVL, is therefore tightly connected with departments at the hospital where data is collected and decisions are made.
The project involves many researchers in Bergen, both clinical and methodological, in addition to national and international collaborators from world-class research institutions in the USA (Mayo Clinic), Switzerland (ETH), Germany (Zuse Institute), France (ISIMA), Luxembourg (LIH) and Poland (TUL).
We aim to recruit new researchers, offer new interdisciplinary courses for medical students, engineers and natural science students, and disseminate and discuss methods and results with a wider audience. There’s also a substantial innovation potential for machine learning in medicine, and the project will be able to identify and potentially pursue such opportunities.
The main goal of the project is to contribute to increased degree of personalized medicine and better decision support for diagnosis, prognosis and therapy in diseases and conditions where images are an important source of information.