AI system Hetairos classifies brain tumors in minutes, outperforming experts

Researchers at the German Cancer Research Center (DKFZ) and Heidelberg University Hospital have created an artificial‑intelligence tool called Hetairos that can determine the molecular subtype of central‑nervous‑system tumours from routine histological slides. Trained on more than 11,000 digitised tissue sections from 9,606 patients across four continents, the system distinguishes 102 molecular subtypes and delivers a diagnosis in roughly 12 minutes, compared with DNA‑methylation tests that can take up to two weeks.

In a head‑to‑head test, Hetairos correctly identified the tumour subtype in 68 % of 210 cases, while a panel of five experienced neuropathologists achieved about 30 % accuracy. When the three most likely diagnoses were considered, the AI’s accuracy rose to 84 % versus 50 % for the human experts. The tool provides a confidence estimate for each case and is intended to support, not replace, pathologists, especially in settings lacking specialised molecular‑diagnostic facilities. Rare tumour types remain a challenge for the algorithm.

By using standard stained slides that are widely available, Hetairos could accelerate and broaden access to precise brain‑tumour classification, helping clinicians prioritise further testing and potentially improving treatment decisions worldwide.