A new artificial intelligence system called "Hetairos" can predict the molecular classification of brain and spinal cord tumors in just twelve minutes using routine tissue samples—a task that typically takes two weeks with current gold-standard methods. In a head-to-head test, the AI outperformed five expert neuropathologists, and its accuracy soared when it expressed high confidence. This breakthrough could transform brain tumor diagnosis, especially in regions lacking specialized labs.
The Research: AI Masters Tumor Subtyping
The international team, led by Moritz Gerstung at the German Cancer Research Center (DKFZ) and Felix Sahm at Heidelberg University, developed the AI by training it on a massive dataset: over 11,000 digitized tissue sections from 9,606 patients across four continents. The ground truth for each sample was established using DNA methylation analysis—the current gold standard for classifying central nervous system tumors.
Hetairos can distinguish between 102 distinct molecular subtypes, covering nearly the entire WHO classification for brain tumors. But its real test came in a clinical trial with 210 complex cases. When pitted against five senior neuropathologists, Hetairos achieved a diagnostic accuracy of 68%, while the humans averaged just 30%. When allowing the AI to list its top three guesses, its accuracy jumped to 84%—more than double the human experts' 50%.
The AI also includes a self-evaluation feature. In 50-70% of cases, it flagged its own predictions as "high certainty," and in those instances, its accuracy reached 87-88%. Moreover, the AI isn't a "black box": it highlights the exact microscopic regions that led to its decision, allowing pathologists to verify its reasoning and even guide targeted follow-up testing.
Why It Matters for Your Brain
Brain tumors are notoriously diverse, and precise molecular subtyping is crucial for choosing the best treatment. Traditional DNA methylation testing requires specialized labs, expensive equipment, and large tissue samples—a luxury not available in many parts of the world. By using standard stained slides, Hetairos could make accurate, rapid diagnosis accessible almost anywhere, potentially reducing the diagnostic delay from twelve days to twelve minutes.
For you, this means that if you ever face a brain tumor diagnosis, AI tools like Hetairos may soon enable quicker, more precise decision-making—potentially improving outcomes. It also illustrates how AI can augment human expertise, not replace it, by handling massive pattern recognition tasks while pathologists focus on complex cases.
What You Can Do
While you can't directly harness this AI, you can stay informed about AI developments in healthcare—they may soon impact diagnosis and treatment. To keep your own brain sharp, engage in varied cognitive challenges like puzzles, memory games, and learning new skills. These activities support neuroplasticity and may help maintain cognitive health as you age.
Source: Neuroscience News
Curious about your own brain? Take our free adaptive IQ test or try 306 brain training levels.