Computer Vision News - January 2024

models in the medical domain (figure below). It can be utilized as a pre-processing step to enhance collateral evaluation. 27 Mumu Aktar Computer Vision News The results of the research led to novel automated collateral evaluation methods, effectively addressing rater variability as well as small and imbalanced datasets. Unlike existing approaches, Mumu’s innovative techniques identify collateral scores from ischemic damage in NCCT or 4D CTA, demonstrating robustness with limited data through transfer learning. Siamese networks expand the methodology to similaritybased problem-solving with minimal data requirements. Congrats, doctor Mumu!