When a patient shows signs of a brain tumor—such as double vision—the usual next step is a biopsy to collect tissue and analyze it. But in some cases, like when the tumor is located in a risky area of the brain, surgery could be too dangerous.
That’s why researchers at Charité—Universitätsmedizin Berlin have created a new method to diagnose brain tumors using artificial intelligence (AI), without needing to operate. Their results were recently published in Nature Cancer.
Instead of relying on a tissue sample viewed under a microscope, the new approach looks at the tumor’s genetic material—specifically, its epigenetic fingerprint. This is a pattern of chemical changes to the DNA that controls which genes are turned on or off. Every cell has these patterns, and they differ from one type of tumor to another.
Even better, this fingerprint can sometimes be found in body fluids like cerebrospinal fluid, which can be collected safely using a small needle in the lower back—a far less risky method than brain surgery.
Using this idea, the researchers built an AI model called crossNN. It’s a type of neural network—a basic AI system trained to recognize patterns. The model was taught to spot the differences between thousands of tumor types by analyzing their epigenetic fingerprints.
It was trained using a large collection of known tumor data and then tested on over 5,000 tumor samples. The model correctly diagnosed brain tumors in 99.1% of cases. It also accurately identified more than 170 tumor types from various parts of the body with 97.8% accuracy.
What makes crossNN special is that it works well even when the DNA information is incomplete or collected using different methods. In the past, this inconsistency made it harder to use AI for diagnosis. But crossNN handles these differences effectively, thanks to its smart design and careful training.
Importantly, the model is not a black box—it’s explainable. This means doctors can understand how the AI reached its diagnosis, which is essential for using it in hospitals. It can work with DNA samples from either tissue or fluids, giving it great flexibility in real-world settings.
The research team already used this technology in a real case. A patient with double vision had a tumor in a location too risky to biopsy.
Instead, doctors collected cerebrospinal fluid and analyzed it with nanopore sequencing, a fast method of reading genetic data. The AI identified the tumor as a lymphoma of the central nervous system—a diagnosis that allowed doctors to start chemotherapy right away, without delay or surgery.
This breakthrough has major implications for the future of cancer care. Traditionally, doctors need a physical sample of the tumor to make a precise diagnosis and choose the right treatment. But for hard-to-reach tumors, this isn’t always possible. The new method could change that.
Because each tumor behaves differently, knowing its exact type is key to choosing the right treatment. Some therapies are highly targeted, working only on certain types of cells. Others require different doses depending on the tumor’s growth rate or location. Fast, accurate diagnosis means treatment can begin sooner and with better results.
The next step for the research team is to run clinical trials across Germany, in partnership with the German Cancer Consortium. They’re also exploring whether this method can be used during surgery to give doctors real-time information. The ultimate goal is to make this safer, faster, and more accurate way of diagnosing cancer a regular part of medical care.
In conclusion, this AI model doesn’t just improve how we diagnose tumors—it changes what’s possible. By replacing risky biopsies with fluid-based tests and using cutting-edge AI to read the tumor’s “fingerprint,” doctors can deliver faster, safer, and more precise care. For patients facing a scary diagnosis, that could make all the difference.
If you care about cancer risk, please read studies that exercise may stop cancer in its tracks, and vitamin D can cut cancer death risk.
For more health information, please see recent studies that yogurt and high-fiber diet may cut lung cancer risk, and results showing that new cancer treatment may reawaken the immune system.
The research findings can be found in Nature Cancer.
Copyright © 2025 Knowridge Science Report. All rights reserved.
