Minnelab at MICCAI 2025

This year, some members of MINNELab had the pleasure of attending MICCAI 2025 in Daejeon, South Korea — an incredible experience to reconnect with the vibrant and inspiring MICCAI community. The conference provided a fantastic opportunity for discussions, collaborations, and the exchange of ideas across medical imaging and AI.

Simone Bendazzoli presented at two workshops:

  • BraTS Workshop — BraTS-FL: Enhancing Generalization in Brain Tumor Segmentation via Federated Learning This work introduced a federated learning approach to improve generalization across brain tumor subtypes, including adult glioma, metastasis, meningioma, pediatric glioma, and sub-Saharan African cohorts. The paper will appear soon in the workshop proceedings.
    The method was also awarded first prize in the BraTS GoaT Task 7 subchallenge on generalizability across tumor types — a rewarding recognition for the effort toward creating more robust and inclusive AI models in neuroimaging.

  • DeCaF (Distributed and Collaborative Federated Learning) 2025 Workshop — MONet-FL: Extending nnU-Net with MONAI for Clinical Federated Learning
    Simone presented the MONet Bundle, an open-source framework that integrates nnU-Net into the MONAI ecosystem, enabling its use in federated learning, active learning, and PACS integration.
    The MONet Bundle was showcased in two federated tasks: whole-body PET-CT lymphoma segmentation and multimodal MRI brain tumor segmentation.

Beyond the scientific exchange, the team deeply appreciated the warm hospitality and rich culture of Korea, from the food and traditions to the welcoming atmosphere throughout Daejeon.

We return from MICCAI 2025 energized and inspired — looking forward to reconnecting with the MICCAI community soon for more exciting discussions, collaborations, and innovations in medical imaging AI.