This year’s 25th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2022, was held from September 18th to 22nd 2022 in Singapore. It was the first time for MICCAI conference to be hosted in Southeast Asia. This year, Albert Clèrigues and Valeriia Abramova, two of VICOROB’s PhD students working in medical imaging, traveled to assist the conference in person. The long trip to Singapore (Around 20 hours of travel time in each way) and the jetlag due to the 6 hours in timezone difference was definitely worth it. Valeriia attended the conference as a finalist of an INSTANCE 2022 challenge (Intracranial Hemorrhage Segmentation Challenge on Non-Contrast head CT (NCCT)). The challenge was a part of a wide variety of satellite events, including other challenges, tutorials, and workshops. At the challenge day, being in a top-10 winning teams, she presented her solution and attended the presentations of other winning teams.
The conference was undoubtedly an extremely positive and enriching experience for our students. Many different aspects of the field of medical imaging were covered from different angles during the conference. It was a big pleasure to see many world-leading scientists from wide range of disciplines both online and in-person participating in discussions, share experience and giving ideas for the future research, pushing forward the field of medical imaging. This year, a lot of attention was dedicated to social and ethical aspects of medical imaging and deep learning domain. Several events were organized to raise awareness and widen the representation of women and minority researchers from low-middle income countries. These were done by Women in MICCAI (WiM) and RISE-MICCAI (Reinforce Inclusiveness & diverSity and Empower) committees. The ethical aspect of AI was covered in the workshop Ethical and Philosophical Issues in Medical Imaging (EPIMI) and during one of the keynote speeches. Moreover, one presentation of the main conference covered the topic of carbon footprint of selecting and training of deep learning models.
Workshop room
Poster room
In purely academic terms, there were a wide variety of interests to the scientific community in the medical imaging domain. In general terms, the academic works of this year proposed more specific solutions and AI frameworks for each imaging modality and pathology than in previous years, were more general approaches and solutions were the norm. The most cutting-edge techniques this year were around transformer-based models, which were popular among the works proposed and they covered different applications, like registration, image reconstruction, segmentation, survival prediction, report generation, etc. Interpretability of AI models and uncertainty estimation of their outputs was also a hot topic. Plenty of oral and poster presentations were dedicated to these topics and, moreover, Uncertainty for Safe Utilization of Machine Learning in Medical Imaging (UNSURE) workshop was held apart of main conference and attracted a lot of attendees – so many, that organizers had to put additional chairs for the audience and still many people had to sit on the floor. Another emerging topic was that of distributed or federated learning, which was also covered during presentations, workshops and tutorials. This topic comprises a set of techniques which allow the training of AI models with data from different hospitals without compromising data protection or patient privacy. Domain adaptation or generalization continues to be a very active field, and had many oral and poster presentations, together with 3 workshops. This works propose different solutions to a problem in which, for example, a model trained on images from one scanner brand performs poorly when used on images from different scanner manufacturers.
But not everything was work at MICCAI, the conference committee organized several events for attendees to explore Singapore – sightseeing trip, the trip to university campus and Gala dinner, which was held in Singapore zoo and where we had the opportunity to visit the night safari. Singapore is a great city with rich culture, great transport infrastructure and huge amount of green space.
View of Singapore
