Generative AI and large language models in nuclear medicine: current status and future prospects

被引：2

作者：

Hirata, Kenji ^{[1
]}

Matsui, Yusuke ^{[2
]}

Yamada, Akira ^{[3
]}

Fujioka, Tomoyuki ^{[4
]}

Yanagawa, Masahiro ^{[5
]}

Nakaura, Takeshi ^{[6
]}

Ito, Rintaro ^{[7
]}

Ueda, Daiju ^{[8
]}

Fujita, Shohei ^{[9
,10
]}

Tatsugami, Fuminari ^{[11
]}

Fushimi, Yasutaka ^{[12
]}

Tsuboyama, Takahiro ^{[13
]}

Kamagata, Koji ^{[14
]}

Nozaki, Taiki ^{[15
]}

Fujima, Noriyuki ^{[16
]}

Kawamura, Mariko ^{[7
]}

Naganawa, Shinji ^{[7
]}

机构：

[1] Hokkaido Univ, Grad Sch Med, Dept Diagnost Imaging, Kita 15,Nishi 7,Kita Ku, Sapporo, Hokkaido 0608638, Japan

[2] Okayama Univ, Fac Med, Dept Radiol Dent & Pharmaceut Sci, Kita Ku, Okayama, Japan

[3] Shinshu Univ Sch Med, Med Data Sci Course, Matsumoto, Nagano, Japan

[4] Tokyo Med & Dent Univ, Dept Diagnost Radiol, Bunkyo Ku, Tokyo, Japan

[5] Osaka Univ, Grad Sch Med, Dept Radiol, Suita, Osaka, Japan

[6] Kumamoto Univ, Grad Sch Med, Dept Diagnost Radiol, Chuo Ku, Kumamoto, Japan

[7] Nagoya Univ, Grad Sch Med, Dept Radiol, Showa Ku, Nagoya, Japan

[8] Osaka Metropolitan Univ, Grad Sch Med, Dept Artificial Intelligence, Abeno Ku, Osaka, Japan

[9] Univ Tokyo, Grad Sch Med, Dept Radiol, Bunkyo Ku, Tokyo, Japan

[10] Univ Tokyo, Fac Med, Bunkyo Ku, Tokyo, Japan

[11] Hiroshima Univ, Dept Diagnost Radiol, Minami Ku, Hiroshima, Japan

[12] Kyoto Univ, Grad Sch Med, Dept Diagnost Imaging & Nucl Med, Kyoto, Japan

[13] Kobe Univ, Grad Sch Med, Dept Radiol, Chuo Ku, Kobe, Japan

[14] Juntendo Univ, Grad Sch Med, Dept Radiol, Bunkyo Ku, Tokyo, Japan

[15] Keio Univ, Sch Med, Dept Radiol, Shinjuku Ku, Tokyo, Japan

[16] Hokkaido Univ Hosp, Dept Diagnost & Intervent Radiol, Kita Ku, Sapporo, Japan

来源：

ANNALS OF NUCLEAR MEDICINE | 2024年 / 38卷 / 11期

关键词：

Generative AI; Large language model; Report generation; Report structuring; Education; Nuclear medicine; PET; SPECT; CHATGPT;

D O I：

10.1007/s12149-024-01981-x

中图分类号：

R8 [特种医学]; R445 [影像诊断学];

学科分类号：

1002 ; 100207 ; 1009 ;

摘要：

This review explores the potential applications of Large Language Models (LLMs) in nuclear medicine, especially nuclear medicine examinations such as PET and SPECT, reviewing recent advancements in both fields. Despite the rapid adoption of LLMs in various medical specialties, their integration into nuclear medicine has not yet been sufficiently explored. We first discuss the latest developments in nuclear medicine, including new radiopharmaceuticals, imaging techniques, and clinical applications. We then analyze how LLMs are being utilized in radiology, particularly in report generation, image interpretation, and medical education. We highlight the potential of LLMs to enhance nuclear medicine practices, such as improving report structuring, assisting in diagnosis, and facilitating research. However, challenges remain, including the need for improved reliability, explainability, and bias reduction in LLMs. The review also addresses the ethical considerations and potential limitations of AI in healthcare. In conclusion, LLMs have significant potential to transform existing frameworks in nuclear medicine, making it a critical area for future research and development.

引用

页码：853 / 864

页数：12

共 84 条

[1] Llama 3 Challenges Proprietary State-of-the-Art Large Language Models in Radiology Board-style Examination Questions [J].