Healthcare

Neural network to help doctors simultaneously identify up to 10 pathologies using CT scans

Neural network to help doctors simultaneously identify up to 10 pathologies using CT scans
Photo by Maxim Mishin. Moscow Mayor and Government Press Service
An integrated artificial intelligence service has already helped analyze more than 270,000 medical images.

The integrated artificial intelligence service has offered new opportunities. The neural network can now detect up to 10 pathologies at a time using a CT scan. According to Anastasia Rakova, Deputy Moscow Mayor for Social Development, adrenal tumors are among the diseases that algorithms identify using chest CT scans.

“The development of AI-based integrated solutions is one of the major experimental areas to introduce CV technologies in diagnostic radiology this year. We already have two integrated tools to detect several pathologies at once using one image. Recently, the neural network has learned to recognize adrenal tumors on chest CT scans. So, one service already can detect up to 10 pathologies on one image. It helps diagnose pathologies in patients who see a doctor to treat other diseases. Since the launch of the integrated artificial intelligence service, it has helped doctors analyze more than 270,000 medical images,” Anastasia Rakova said.

In addition to adrenal tumors, AI can detect signs of lung cancer, COVID-19, spinal osteoporosis, thoracic aortic aneurysm, pulmonary hypertension, hydrothorax, pulmonary emphysema, coronary heart disease all at once by the degree of the coronary arteries calcification and the volume of paracardial fat present on CT scans.

Algorithms highlight potentially abnormal areas on medical images with color clues, automatically measure them, and generate descriptions thereof required to make a diagnosis. CV technology is used to identify abnormalities at an early stage when they can be invisible to the human eye.

Moscow has been digitizing its healthcare system for over 10 years. Since 2020, smart algorithms have been helping doctors find abnormalities using radiology images as part of a project to introduce CV technologies into Moscow healthcare institutions.

Automation based on artificial intelligence technologies increases productivity and quality of work of radiology specialists and renders healthcare more accessible for their patients. At the same time, the project facilitates development of the artificial intelligence services market in the healthcare sector. It is supported by the Center for Diagnostics and Telemedicine of the Moscow Healthcare Department in liaison with the Department of Information Technologies.