Tomikawa Emi, Sakai Satoshi, Yamagiwa Yoshinori, Kangawa Yumi, Kagawa Yusuke, Kato Yuki, Kojima Kensuke, Inomata Akira, Matsumoto Izumi, Sato Gen, Suzuki Mutsumi
AI Pathology Task Force, Non-clinical Evaluation Expert Committee, Drug Evaluation Committee, Japan Pharmaceuticals Manufacturers Association, 2-3-11 Nihonbashi Honcho, Chuo-ku, Tokyo 103-0023, Japan.
Toray Industries, Inc., 6-10-1 Tebiro, Kamakura-shi, Kanagawa 248-8555, Japan.
J Toxicol Pathol. 2025 Jul;38(3):191-198. doi: 10.1293/tox.2024-0100. Epub 2025 Mar 11.
The use of artificial intelligence (AI) in non-clinical pathology is rapidly expanding. In this study, we conducted a literature survey of articles published after 2017 that used AI to analyze the histopathological images of experimental animals. We identified 44 articles that used AI for various purposes, including the detection of abnormal sites, determination and quantification of normal tissues, and classification of normal/abnormal images. AI systems or applications were either custom-built, commercially available, or a combination of both. Rats and mice were mainly used, and the liver was the most frequently analyzed organ. Our findings suggest that AI can be useful in non-clinical pathology and that collaboration between pharmaceutical companies or cooperation with IT experts can be a potential approach to further advance the utilization of AI in this field.
人工智能（AI）在非临床病理学中的应用正在迅速扩展。在本研究中，我们对2017年后发表的使用AI分析实验动物组织病理学图像的文章进行了文献调查。我们识别出44篇将AI用于各种目的的文章，包括异常部位的检测、正常组织的测定和定量，以及正常/异常图像的分类。AI系统或应用程序要么是定制的，要么是商业可用的，或者两者兼而有之。主要使用大鼠和小鼠，肝脏是分析最频繁的器官。我们的研究结果表明，AI在非临床病理学中可能有用，制药公司之间的合作或与IT专家的合作可能是进一步推进该领域AI应用的潜在途径。
