Japan-Medical Image Database (J-MID): Medical Big Data Supporting Data Science

The digitization of radiology practices has advanced for nearly two decades, exemplified by the global standard Digital Imaging and Communications in Medicine (DICOM). Rapid technological progress in imaging modalities has led to a significant increase in the volume of data handled. However, it has become difficult for the limited number of radiologists in Japan to maintain the quality of diagnosis while efficiently processing the data. In response, the Japan Radiological Society (JRS) advocated the "Japan Safe Radiology" Initiative, which aims to improve the safety, efficiency, and quality of radiological medicine by actively utilizing information and communication technology (ICT) in all aspects of radiological practice. Recent advances in innovative artificial intelligence (AI) technology have shown a high affinity for image processing, prompting recognition of the importance of using big data systems to integrate radiological medicine. Consequently, in 2017, the Japan Agency for Medical Research and Development (AMED) supported the JRS project, Development Research for the Realization of a National Image Diagnosis Database, through which the Japan Medical Image Database (J-MID) was established. The J-MID is designed to centralize medical resources and systematically collect CT/MR images and diagnostic reports from 10 major university hospitals in Japan through an academic information network (SINET). Data were anonymized and stored on a central server in the cloud, enabling researchers to utilize J-MID conveniently. In April 2024, J-MID had collected more than 534 million images (1.65 million cases), making it an unparalleled repository of real-world radiological data in Japan.