Introduction to Clinical Trials and Case Reports for Medical Image AI Products

Despite multiple impacts and disruptions caused by the COVID-19 pandemic in 2020, the number of clinical trials and certificates obtained from AI intelligent-assisted diagnostic products based on medical image data with large quantities and deep learning technology continues to grow steadily. To date, more than 40 deep-learning-based AI-independent software Class III products have been approved and listed by the National Medical Products Administration (NMPA) in China; among them, the main products are image software Class III, especially computed tomography (CT) image products, which account for more than 95% of the total. At present, the medical image AI products that have been marketed worldwide can be divided into the following categories according to their clinical functions: (1) AI-assisted clinical decision-making (e.g. detecting lesion location and determining benignity or malignancy); (2) AI-based pre-processing for improving image quality, enhancing image speed and image reconstruction; (3) AI optimization of clinical workflows; and (4) AI-based post-processing for image segmentation and measurement analysis.