In this review, methodologies of Kansei Engineering have been introduced. Kansei Engineering consists of techniques for measuring Kansei and analyzing it mathematically, then applying the results to product development. The basic methods of KEs are modified semantic differential method, principal component analysis and multiple regression. KE has been studied to expand the boundaries of the methods used to analyze Kansei. Machine learning methods including neural networks have been developed and applied. Other mathematical methods such as morphometrics are also applied. Measurement based 3D CG, VR are used for detailed expression of model details. Finally, two AI based challenges are presented. One is CNN based Kansei mimicking system and another is generative AI for exploring novel design ideas based on KE.

Generative AI has garnered societal attention, with AI usage advancing in various settings. It has become involved in situations ranging from creative domains to everyday tasks, such as idea generation, source code writing, and text summarization. The broad applicability of generative AI is distinct from the period before generative AI became a topic of discussion. Notable services utilizing generative AI include image generation services like OpenAI’s DALL·E 3 and Stability AI’s Stable Diffusion. In text generation, well-known services are OpenAI’s ChatGPT and Google’s Bard. The rise of such generative AIs is underpinned by the emergence of new deep-learning models and changes in training method of models. This shift can alter the AI development framework in medical image processing. This paper provides an overview of diffusion models and large language models used in generative AIs and introduces examples of their application in medical image processing. It also discusses foundation models that have garnered attention about generative AI and mentions a new AI development style using foundation models. Furthermore, this paper explores methods to develop domestically produced foundation models and high-performance AIs for medical image processing that are internationally viable while maximizing the use of Japan’s information and computing resources.

The novel coronavirus indicates the potential occurrence of future pandemics of unknown viruses. Contact infection prevention in hospitals can exhaust medical staff, resulting in inappropriate medical care. We developed a noncontact operation system using the motion sensor of the radiographic console attached to the flat panel detector to prevent contact infection in preparation for potential unknown viruses in the future. This study investigated its usability and feasibility for clinical use. Experiments in which a radiographic console was simulated using the noncontact operation system developed in this study and a conventional trackpad method were conducted, and the average time from the beginning to the end of the operation was determined. The intraoperator comparison showed that operation using the system developed in this study was significantly faster than the conventional trackpad method (20.9 ± 5.3 vs. 25.5 ± 7.6 s, P < 0.01). Additionally, the interoperation comparison reported that the operation method using this system was significantly faster than the conventional trackpad method (17.5 ± 3.8 vs. 25.5 ± 7.6 s, P < 0.01). The system developed in this study can be used to perform noncontact operations faster than conventional methods. Therefore, the system can potentially be used for controlling infection.