We are delighted to announce that RadioVal will be represented at the upcoming MICCAI 2025 main conference, the leading international venue for medical image computing and computer-assisted interventions.
Main Conference Paper
Our team will present the paper:
This work addresses one of the pressing challenges in medical AI: robust generalization across diverse imaging settings. Standard models often struggle when faced with distribution shifts in clinical data. Our proposed method introduces a multi-view co-training approach for single-image test-time adaptation, enabling models to adjust to new domains using only the information available at inference.
The approach significantly improves reliability in real-world deployment scenarios, where large batches of adaptation data are rarely available.
RadioVal at MICCAI 2025
This contribution underscores RadioVal’s mission to ensure that medical AI is not only accurate in controlled research settings but also robust, adaptable, and clinically relevant across diverse healthcare environments.
We look forward to sharing our results with the community at MICCAI 2025 and to further discussions on strengthening the resilience of medical imaging AI.