UNI and CONCH (Harvard, Mahmood Lab)
Two open academic pathology models: UNI reads tissue images, CONCH links images with pathology text.
UNI (Nature Medicine 2024) was pretrained on 100M tiles from 100k slides; UNI2-h scaled further. CONCH (Nature Medicine 2024) is a vision-language model trained on 1.17M image-caption pairs enabling zero-shot classification and retrieval. Both released for research and widely used as baselines.
How it works
UNI is trained with DINOv2 self-supervision; CONCH uses contrastive image-text alignment.
- Open weights for research
- Reproducible baselines
- Non-commercial licence
- Tile-level, needs aggregation
Latest papers
topQuery for this technology: (TITLE:"UNI and CONCH" OR ABSTRACT:"UNI and CONCH" OR TITLE:"Harvard, Mahmood Lab" OR ABSTRACT:"Harvard, Mahmood Lab") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about UNI and CONCH (Harvard, Mahmood Lab), not a curated reading list.