A neural network was trained to detect 10 pathologies on a single CT scan
In Moscow, a neural network learned to simultaneously identify up to 10 pathologies on a CT scan. In addition to heart and lung diseases, artificial intelligence now detects adrenal gland diseases.Developed for Moscow hospitals neural network can simultaneously detect up to 10 pathologies on a computer tomogram (CT) of the chest, the press service of the Department of Health of the capital. The diseases that the algorithms help doctors identify include adrenal gland masses. It is planned that this will make it possible to diagnose pathologies in patients who have turned to a doctor for treatment of other diseases.
In addition to adrenal masses, the artificial intelligence (AI) can simultaneously find signs of lung cancer, COVID-19, spinal osteoporosis, thoracic aortic aneurysm, pulmonary hypertension, hydrothorax, pulmonary emphysema, and coronary heart disease on a CT scan by the degree of coronary artery calcification and the volume of paracardial fat.
The algorithms mark areas of possible pathologies with color cues on the medical image. A neural network can detect pathologies at an early stage, when they may not be visible to the doctor.
Despite the fact that half of Russians believe that artificial intelligence can be used in diagnosis and treatment, only one in six is ready to entrust the neural network with their own health. At the same time, three quarters of the population would not want such technology to replace a doctor. At the same time, the authorities have embarked on a course to robotize medicine: in early 2023, the Ministry of Health instructed the regions to urgently implement at least one AI-based solution at medical institutions. President Vladimir Putin last autumn demanded that special attention be paid to the implementation of such technologies in healthcare, especially in primary care.