Duan Chong Ling, An Lin, Yang Yong Feng, Yuan Lili, Zhu Yandong, Han Qian, Ma Hongbing, Zhao Fei, Yu Qing-Qing
Jining No.1 People's Hospital, Shandong First Medical University, Jining, People's Republic of China.
Clinical Research Center, Jining No. 1 People's Hospital, Shandong First Medical University, Jining, People's Republic of China.
Cancer Manag Res. 2025 Jul 19;17:1457-1475. doi: 10.2147/CMAR.S529589. eCollection 2025.
Lymphomas are a hematopoietic malignancies that encompass over 90 subtypes. Traditionally, they have been categorized into two main groups, non-Hodgkin lymphoma (NHL) and Hodgkin lymphoma (HL). Based on morphology and immunohistochemistry, HL can be classified into nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL) and classical HL (cHL). NHL represents the most common form of lymphoma, including more than 50 subtypes, such as mantle cell lymphoma (MCL), follicular lymphoma (FL), marginal zone lymphoma (MZL), and the most common, diffuse large B-cell lymphoma (DLBCL). Medical imaging plays a pivotal role in lymphoma management, with positron emission tomography/computed tomography (PET/CT) serving as an indispensable tool. 2-Deoxy-2-[fluorine-18]fluoro-D-glucose (18F-FDG) PET/CT is extensively utilized in lymphoma management, having demonstrated its value in providing crucial data for precise disease burden quantification, treatment response evaluation, and prognostic assessment. Radiomics is an innovative approach that entails the computer-aided extraction of quantitative, searchable data from medical images and its association with biological and clinical outcomes. The rapid advancement of radiomics research has opened new avenues for cancer diagnosis and therapy. Our findings indicate that artificial intelligence based PET/CT radiomics has demonstrated significant potential in lymphoma diagnosis, subtyping, staging, treatment selection, and survival prognosis assessment, offering clinicians powerful decision-support tools. However, challenges remain, such as the lack of standardized image quality in machine learning applications.
淋巴瘤是一种造血系统恶性肿瘤，包含90多种亚型。传统上，它们被分为两个主要类别，非霍奇金淋巴瘤（NHL）和霍奇金淋巴瘤（HL）。根据形态学和免疫组织化学，HL可分为结节性淋巴细胞为主型霍奇金淋巴瘤（NLPHL）和经典型HL（cHL）。NHL是淋巴瘤最常见的形式，包括50多种亚型，如套细胞淋巴瘤（MCL）、滤泡性淋巴瘤（FL）、边缘区淋巴瘤（MZL），以及最常见的弥漫性大B细胞淋巴瘤（DLBCL）。医学成像在淋巴瘤管理中起着关键作用，正电子发射断层扫描/计算机断层扫描（PET/CT）是不可或缺的工具。2-脱氧-2-[氟-18]氟-D-葡萄糖（18F-FDG）PET/CT在淋巴瘤管理中被广泛应用，已证明其在提供精确疾病负担量化、治疗反应评估和预后评估的关键数据方面的价值。放射组学是一种创新方法，需要从医学图像中计算机辅助提取定量的、可搜索的数据，并将其与生物学和临床结果相关联。放射组学研究的快速发展为癌症诊断和治疗开辟了新途径。我们的研究结果表明，基于人工智能的PET/CT放射组学在淋巴瘤诊断、亚型分类、分期、治疗选择和生存预后评估方面已显示出巨大潜力，为临床医生提供了强大的决策支持工具。然而，挑战依然存在，比如机器学习应用中缺乏标准化的图像质量。
