The award recognises outstanding work in the field of Artificial Intelligence and promotes promising early-career researchers in Austria. Her thesis, conducted at Computational Imaging Research Lab under the supervision of Philipp Seeböck, investigates deep learning-based methods for identifying regions of breast tissue at risk of developing suspicious lesions in DCE-MRI scans of high-risk patients.
Breast cancer remains the most common cancer in women, and early detection is crucial, especially in high-risk groups. A new master’s thesis from our lab explored how deep learning can support risk assessment using breast DCE-MRI scans. In the course of the master thesis Bettina Röthlin developed a segmentation pipeline to identify regions in MRI scans that are more likely to develop lesions within the following 6–24 months. Comparing different architectures. Features derived from the resulting probability maps proved informative for classifying scans as higher or lower risk, reaching an accuracy of 61% and precision of 68%. While further refinement is needed, the work demonstrates the potential of integrating segmentation-based features into risk stratification pipelines for high-risk breast cancer patients, and might be relevant for future adaptive screening paradigms incorporating updated risk profiles of patients.