Fujifilm and the National Cancer Center have jointly developed an "AI Development Support Platform," which will help doctors and researchers develop AI technologies (software) for diagnostic imaging assistance without requiring programming or other expertise. The National Cancer Center will conduct several studies on various research subjects with the application of this platform to verify its usefulness, and Fujifilm will work towards commercialization.
To develop diagnostic imaging support AI technology, it is common to use multiple tools to create a large amount of data to have the AI learn, design a training model to determine the best learning method, and then implement the training. However, the annotation tools that are currently widely used for processing, such as those giving correct information to image data, are not optimized for medical images, and subsequent learning processes also need to utilize individual tools, and advanced engineering knowledge is necessary to execute the series of development processes. The National Cancer Center has been engaged in the "Development of an Integrated Cancer Medical System Using Artificial Intelligence" project since 2016 and the "Development of an Innovative Cancer Drug Discovery System Using Artificial Intelligence" project since 2018 as part of the strategic research promotion project by the Japan Science and Technology Agency (JST) in collaboration with Chuo Hospital. This "AI Development Support Platform" was constructed as an integrated development environment for AI technology for imaging diagnostics assistance through collaboration between the National Cancer Center and Fujifilm, using Fujifilm’s knowledge of advanced image editing in diagnostics imaging systems and AI development technology for 3-D images that take full advantage of DGX performance, based on the results of the annotation platform for clinical data structurization obtained in these projects. This platform provides various support functions that doctors can utilize for research and development of clinical support software technologies using AI technology.
The development of diagnostic support technologies for medical images requires repeated processes such as machining (annotation) and management of a large amount of training data, setting up training models, and the execution of training. In contrast, the developed platform provides an environment that enables the performing of a series of AI development processes from the creation of training data to the execution and evaluation of learning, to be executed even without advanced engineering knowledge such as efficient and intuitive image viewing and annotation, with an operational feel similar to that of the diagnostics imaging environment used in clinical settings. The development of this platform is expected to further accelerate the research and development of image diagnostics support technology using AI technology.
This article has been translated by JST with permission from The Science News Ltd.(https://sci-news.co.jp/). Unauthorized reproduction of the article and photographs is prohibited.