Moscow has updated the digital data library for testing neural networks in healthcare and increased the number of open datasets to 40.

Previously, developers of AI-powered medical services had access to only nine datasets. Now the datasets cover most of the main medical tests: ultrasound, ECG, CT, MRI, X-ray examinations and others. Using the datasets, developers will be able to train their own models to analyze medical images for pathologies.

Currently, the most popular dataset contains chest CT images for training AI-based services to detect the signs of COVID-19. It has been viewed over 4,000 times and downloaded more than 700 times. The second one is the combined dataset for several types of medical tests, such as CT, mammography, X-rays and fluorography. The third most popular dataset is the brain MRI dataset. Most of the datasets are used for calibration, functional and other types of testing AI services.

Original in Russian