Nuclear Medicine Unit, A.R.N.A.S Ospedale Civico Di Cristina e Benfratelli, Palermo 90127, Italy.
Division of Nuclear Medicine, Department of Medical Sciences, AOU Città della Salute e della Scienza, University of Turin, Turin 10126, Italy.
Curr Med Imaging. 2023;19(8):832-843. doi: 10.2174/1573405619666230126093806.
F-FDG PET/CT imaging represents the most important functional imaging method in oncology. European Society of Medical Oncology and the National Comprehensive Cancer Network guidelines defined a crucial role of F-FDG PET/CT imaging for local/locally advanced breast cancer. The application of artificial intelligence on PET images might potentially contributes in the field of precision medicine.
This review aims to summarize the clinical indications and limitations of PET imaging for comprehensive artificial intelligence in relation to breast cancer subtype, hormone receptor status, proliferation rate, and lymphonodal (LN)/distant metastatic spread, based on recent literature.
A literature search of the Pubmed/Scopus/Google Scholar/Cochrane/EMBASE databases was carried out, searching for articles on the use of artificial intelligence and PET in breast tumors. The search was updated from January 2010 to October 2021 and was limited to original articles published in English and about humans. A combination of the search terms "artificial intelligence", "breast cancer", "breast tumor", "PET", "Positron emission tomography", "PET/CT", "PET/MRI", "radiomic"," texture analysis", "machine learning", "deep learning" was used.
Twenty-three articles were selected following the PRISMA criteria from 139 records obtained from the Pubmed/Scopus/Google Scholar/Cochrane/EMBASE databases according to our research strategy. The QUADAS of 30 full-text articles assessed reported seven articles that were excluded for not being relevant to population and outcomes and/or for lower level of evidence. The majority of papers were at low risk of bias and applicability. The articles were divided per topic, such as the value of PET in the staging and re-staging of breast cancer patients, including new radiopharmaceuticals and simultaneous PET/MRI.
Despite the current role of AI in this field remains still undefined, several applications for PET/CT imaging are under development, with some preliminary interesting results particularly focused on the staging phase that might be clinically translated after further validation studies.
正电子发射断层扫描/计算机断层扫描（PET/CT）成像代表了肿瘤学中最重要的功能成像方法。欧洲肿瘤内科学会和美国国家综合癌症网络指南定义了 F-FDG PET/CT 成像在局部/局部晚期乳腺癌中的关键作用。人工智能在 PET 图像上的应用可能会在精准医学领域发挥作用。
本综述旨在根据最新文献，总结基于人工智能的 PET 成像在乳腺癌亚型、激素受体状态、增殖率、淋巴结/远处转移扩散方面的临床适应证和局限性。
对 Pubmed/Scopus/Google Scholar/Cochrane/EMBASE 数据库进行文献检索，检索关于人工智能和乳腺癌中 PET 使用的文章。检索从 2010 年 1 月更新到 2021 年 10 月，仅限于以英语发表的关于人类的原始文章。使用的搜索词组合包括“人工智能”“乳腺癌”“乳腺肿瘤”“PET”“正电子发射断层扫描”“PET/CT”“PET/MRI”“放射组学”“纹理分析”“机器学习”“深度学习”。
根据我们的研究策略，从 Pubmed/Scopus/Google Scholar/Cochrane/EMBASE 数据库中获得的 139 条记录中，按照 PRISMA 标准，选择了 23 篇文章。对 30 篇全文文章的 QUADAS 评估报告了 7 篇因与人群和结果无关和/或证据水平较低而被排除的文章。大多数文章的偏倚和适用性风险较低。文章按主题分类，例如 PET 在乳腺癌患者分期和再分期中的价值，包括新的放射性药物和同时进行的 PET/MRI。
尽管人工智能在这一领域的当前作用仍未确定，但 PET/CT 成像的一些应用正在开发中，一些初步的有趣结果特别集中在分期阶段，在进一步验证研究后可能会在临床上得到转化。
