Professor Irina Voiculescu and colleagues recognised at MIUA 2025 for work on medical image segmentation

A research collaboration between the American University of Armenia and the University of Oxford has produced results in the area of segmenting out organs from medical images.

Professor Irina Voiculescu and co-authors alumna Varduhi Yeghiazaryan and her graduate student Yeva Gabrielyan (American University of Armenia) have presented the paper 'Semantic Segmentation with Spreading Scribbles' at the Conference on Medical Image Understanding and Analysis (MIUA). Their paper was selected as the runner-up for the Best Paper Award.

Even when a clinician draws only a rough scribble inside an organ, rather than annotating images with carefully drawn contours, the algorithm can still infer where the organ is. The paper shows how the method takes such scribbles and automatically segments the myocardium and ventricles from cardiac MRI.

The method generalises to other medical images, hence reducing the burden on clinical annotators while maintaining high segmentation quality across diverse datasets.