AI Cancer Detection: 170+ Types Identified
- Faced with the challenge of diagnosing a brain tumor in a patient experiencing double vision, researchers at Charité—Universitätsmedizin Berlin developed an artificial intelligence model to improve tumor diagnostics.
- The findings, published in Nature Cancer, highlight the modelS ability to distinguish between more than 170 diffrent cancer types.
- Tumors are now classified by molecular characteristics, not just the organs they originate in.
AI Model crossNN Excels at Tumor Diagnostics, Cancer Type Detection
Updated June 8, 2025
Faced with the challenge of diagnosing a brain tumor in a patient experiencing double vision, researchers at Charité—Universitätsmedizin Berlin developed an artificial intelligence model to improve tumor diagnostics. The AI, dubbed crossNN, analyzes the epigenetic fingerprint of tumors to classify them quickly and reliably.
The AI model, crossNN compares the epigenetic data of unknown tumors to the fingerprints of over 8,000 reference tumors. A cross marks the tumor to be examined. The visualization shows the large amount of data on which the model is based. Each dot represents the profile of a reference tumor containing several hundred thousand pieces of information; each color represents a specific tumor type. Charité | Philipp Euskirchen. Credit: Charité | Philipp Euskirchen
The findings, published in Nature Cancer, highlight the modelS ability to distinguish between more than 170 diffrent cancer types. This is crucial because effective treatment hinges on accurate cancer type detection.
Tumors are now classified by molecular characteristics, not just the organs they originate in. Targeted therapies, selected chemotherapies and innovative study-based treatments depend on knowing the precise tumor type.
According to Dongsheng Yuan et al,crossNN compares a tumor’s epigenetic data to over 8,000 reference tumors.
What’s next
As targeted therapies evolve, crossNN could help doctors choose treatments and tailor dosages, especially for rare cancers.