Deep Learning for Extracting Tree Structures in Medical Images
Machine learning methods have become an indispensable tool for classification problems in medical image analysis. However, problems involving structured outputs, such as automatically extracting the coronary artery tree in a CT image, remain challenging and learning methods used for classification cannot directly be applied. In this project, we will develop methods and theory for robustly computing structured outputs based on deep learning in combination with shape modelling.
The project is in collaboration with the SCAPIS study, a population study with collected CT examinations in over 28,000 individuals. These examinations can be used to analyze possible atherosclerosis in the coronary arteries of the heart. The size and appearance of atherosclerosis are important biomarkers and good predictors for determining the risk of getting a myocardial infarction in the future. Training data from the SCAPIS study will consist of over 600 manually annotated cases.
Fredrik Kahl (contact)
Full Professor at Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering, Imaging and Image Analysis
Post doc at Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering, Imaging and Image Analysis
Sahlgrenska University Hospital
Chalmers AI Research Centre (CHAIR)
Funding Chalmers participation during 2020–2021
Related Areas of Advance and Infrastructure
Information and Communication Technology
Areas of Advance