Russian Scientists Train AI to Detect Breast Cancer at Early Stages
Researchers in St. Petersburg have developed an AI system that flags early signs of breast cancer on CT scans, reducing diagnostic errors and eliminating the need for repeat testing.
Scientists from Saint Petersburg Electrotechnical University “LETI” (SPbGETU LETI) together with experts from the Almazov National Medical Research Centre have developed a clinical decision support system called OncoDetect AI.
The team trained a neural network to identify early-stage breast cancer using CT scans. The system was trained on 65 studies and 7,000 anonymized images from the Almazov National Medical Research Centre database. Each study contained breast cancer lesions, though not every scan slice showed tumors. In testing, the AI model covered the core tasks of breast cancer diagnosis and workflow management.
Faster and More Accurate
The developers say the system improves both the speed and accuracy of examinations while removing the need for repeat diagnostics.
