44 ECCV Workshop Israel, Germany, Belgium, Italy, Hong Kong, Korea, Abu Dhabi, among others. Based on the reviews, the program committee awarded the best paper award to the paper: “ Anatomy-Aware Contrastive Representation Learning for Fetal Ultrasound ”, authored by Zeyu Fu and colleagues from the University of Oxford, UK. Thepaper presentshowanovel self-supervised contrastive representation learning approach combining anatomical information to design the latent space can be leveraged to improve downstream tasks such as cross-domain segmentation and classification of fetal ultrasound images. Fetal imaging was also the topic of the runner-up paper by Yael Zaffrani-Reznikov from the Technion. Yael’s presentation demonstrated how we can non-invasively assess the functional maturation of fetal lungs by analyzing diffusion-weighted MRI data . A key challenge in such analysis is the inevitable motion of the fetus. To address it, Yael presented qDWI-Morph , a physically-driven self-supervised deep- neural-network approach for simultnaous motion compensation and quantitative diffusion-weighted MRI analysis. Another runner-up paper presented by Alona Golts of the Technion addressed the needfor cell-levelautomaticdetectionand classification required for personalized cancer treatment . Alona presented a novel dataset of Proteasomestained Multiple Myeloma (MM) bone marrow slides , containing nine categories with unique morphological traits. With the relative difficulty of acquiring high-quality annotations in the medical-imaging domain, the dataset was intentionally modes and operational uncertainties before being tested in human clinical trials. Alex demonstrated the ISCT concept by his work on the design of medical devices for vascular aneurysms. The expert panel members: Leo Joskowicz , Hayit Greenspan , Ilan Shelef , Ron Kikinis and William (Sandy) Wells with Tal Arbel discussed the challenges they envisage for medical computer vision from both clinical and technical perspectives. Sharing their multiple-years perceptive in the field, they identified the need for more standard infra-structure for data analysis in hospitals, the need for improved confidence and explainability of deep- learning methods and the need to find large-scale solutions rather than focusing on specific applications. They also pointed out the need of a shared language between clinicians and technological people. The peer-reviewed scientific program composed from 26 full papers, selected from 37 submissions by a rigorous peer-review process with at least three reviewers and an additional meta-reviewer for each paper. The geographic distribution of the submission well represented the international nature of the workshop with submissions from China, India, UK, USA, BEST ECC Hybrid and in-person panel participants.