Symposium on Artificial Intelligence in Medical Imaging in Switzerland
In June, our group took part in the first Symposium on Artificial Intelligence in Medical Imaging (SAIMI) in Bern, co-organized by Prof. Lisa Koch, former head of our group and now leading the Machine Learning in Medicine Lab at the University of Bern.
The symposium featured keynote talks on medical foundation models, uncertainty, reliability, and trustworthiness. Compared to larger venues like MICCAI, SAIMI felt more targeted and closely aligned with our own work, with more room for discussion.
At the poster session we presented five posters, including some early-stage projects:
- Interpretable medical foundation models
- The impact of using large unlabeled cross-sectional datasets for longitudinal modeling
- Interpretable survival analysis for eye disease progression
- Mitigation of shortcut learning
- The effect of training corpora on vision-language model performance.
The session led to a number of valuable discussions with other researchers working on related problems.
Afterwards, we were invited to visit the Machine Learning in Medicine Lab, which develops safe and reliable data science tools with a focus on diabetes care. Talks and short demo sessions gave us insight into their current work, and we look forward to continuing the collaboration between our groups.
The trip also gave us a chance to see Bern in pristine summer weather, with its historic city center and the meandering Aare. We spent time wandering through the city and had a team dinner in the green patio of a lovely Indian restaurant.
Overall, SAIMI 2026 let us present our current work, gather feedback, and reconnect with a group we've long been close to, now working on closely related problems in Bern.
