1305-AI-Driven Subtyping from CNS Tumor Histology

The paper introduces Hetairos, a pioneering artificial intelligence algorithm designed to identify 102 different types of central nervous system tumors using standard histopathology slides. While traditional DNA methylation profiling provides high diagnostic accuracy, it is often hindered by high costs, long wait times, and a need for significant tissue samples. Hetairos addresses these challenges by analyzing digital images of common H&E-stained sections, delivering results in only 12 minutes compared to the typical two-week molecular testing period. Developed using a massive dataset of nearly 10,000 patients across four continents, the model demonstrated superior performance when compared to board-certified neuropathologists. By offering high-confidence predictions and identifying ambiguous cases, this tool serves as a vital resource for global neuro-oncology diagnostics.References:Jin D, Shmatko A, Patel A, et al. Hetairos is a histology-based artificial intelligence model for predicting central nervous system tumor methylation subtypes[J]. Nature Cancer, 2026: 1-15.前往小宇宙评论区与主播互动