AI in CardioVascular Medicine Laboratory (AI-CVM)

Bern University Hospital · Switzerland

We build state-of-the-art machine learning and deep learning for cardiovascular disease. Our work spans multimodal data — medical images across modalities, signals, video, electronic health records, text, and omics — to tackle a broad range of clinical questions.

🌐 Website · 💻 GitHub

What we release here

Open datasets, model weights, and tools for reproducible cardiovascular AI research.

Featured projects

• CCT-FM — Foundation model and the largest comprehensive expert-annotated cardiac CT segmentation dataset (14 structures).
• CTAug — CT-specific augmentation library (metal, motion, calcification, wire artifacts).
• nnUZoo — Unified benchmark of CNN vs. Transformer vs. Mamba for medical image segmentation.
• HolOrama — GUI for cardiac IVUS, OCT, and CT segmentation and editing.
• AAOCA — Detection and classification of anomalous aortic origin of coronary arteries in CCTA.

Get in touch

Interested in collaborating? Visit ai-cvm.org or explore our repositories.