Russian Scientists Train AI to Spot Brain Tumors on MRI Scans

Researchers in Moscow built a web app that detects brain tumors with 97 percent accuracy, a glimpse into how AI is rapidly reshaping medical diagnostics.

A group of young scientists at Sechenov First Moscow State Medical University has developed a web application that can automatically identify brain tumors in MRI scans. The tool goes beyond detection: it pinpoints the exact location of the growth and classifies its type. Researchers say this could boost diagnostic accuracy at early stages — when treatment is most effective — and ease the workload of radiologists.

The system runs on YOLO v11, one of the most advanced neural network models available. To train it, the team fed the model more than 5,000 standardized MRI images. In tests, the algorithm achieved 97.1 percent accuracy in spotting tumors.

Right now, the project exists as a prototype. The researchers built a server-side application with a web interface to demonstrate the trained model in real time. Next steps include expanding the dataset with scans from radiology labs, teaching the AI to detect other brain pathologies, and refining the app with feedback from medical experts.

If further developed, the project could make its way into real-world healthcare settings. The Sechenov University team believes AI-powered tools like this won’t just accelerate diagnostics — they could eventually improve treatment outcomes and boost patients’ chances of recovery.