| 桑睿,王晓庆,李德毅,朱佳琳,刘芮菁,魏玺.影像组学联合多模态超声鉴别良、恶性浅表淋巴结[J].中国医学影像技术,2026,42(6):839~844 |
| 影像组学联合多模态超声鉴别良、恶性浅表淋巴结 |
| Radiomics combined with multimodal ultrasound for differentiating benign and malignant superficial lymph nodes |
| 投稿时间：2025-11-30 修订日期：2026-05-01 |
| DOI：10.13929/j.issn.1003-3289.2026.06.006 |
| 中文关键词: 淋巴结 多模态显像 影像组学 显微超声造影 |
| 英文关键词:lymph nodes multimodal imaging radiomics super resolution contrast-enhanced ultrasound |
| 基金项目:国家自然科学基金(82402274)、天津市教委科研计划项目(2025ZD002)、天津市医学重点学科(专科)建设项目(TJYXZDXK-009A)。 |
| 作者 | 单位 | E-mail | | 桑睿 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | | | 王晓庆 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | | | 李德毅 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | | | 朱佳琳 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | | | 刘芮菁 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | | | 魏玺 | 天津医科大学肿瘤医院超声诊疗科, 国家恶性肿瘤临床医学研究中心, 天津市肿瘤防治重点实验室, 天津市恶性肿瘤临床医学研究中心, 天津 300360 | weixi@tmu.edu.com |
|
| 摘要点击次数: 182 |
| 全文下载次数: 64 |
| 中文摘要: |
| 目的 探讨影像组学联合多模态超声鉴别良、恶性浅表淋巴结的价值。方法 前瞻性收集117例接受多模态超声检查的浅表异常淋巴结患者,以病理为金标准分为良性组(n=33)与恶性组(n=84),同时以8∶2比例划分训练集(n=93)与测试集(n=24)。勾画异常浅表淋巴结ROI以获取其定量参数,并提取其多模态影像组学特征,以logistic回归算法构建多模态超声-临床模型、多模态超声-影像组学模型及其联合模型,利用受试者工作特征曲线评估模型效能。结果 多模态超声-临床模型、多模态超声-影像组学模型及联合模型在训练集的敏感度、特异度、准确率及曲线下面积(AUC)分别为79.83%、76.51%、78.91%及0.826,73.41%、75.01%、69.20%及0.775,以及87.20%、88.19%、84.56%及0.876;在测试集分别为43.82%、100%、60.91%及0.750,88.24%、57.14%、79.17%及0.741,以及87.02%、85.48%、85.70%及 0.875。结论 影像组学联合多模态超声对鉴别良、恶性浅表淋巴结具有良好效能。 |
| 英文摘要: |
| Objective To explore the value of radiomics combined with multimodal ultrasound for differentiating benign and malignant superficial lymph nodes. Methods Totally 117 patients with abnormal superficial lymph nodes who would undergo multimodal ultrasound examinations were prospectively enrolled and divided into benign group (n=33) and malignant group (n=84) based on pathology, also into training set (n=93) and testing set (n=24) at a ratio of 8∶2. ROI of the abnormal superficial lymph nodes were delineated to obtain quantitative parameters, and radiomics features of lymph nodes in multimodal images were extracted. Then multimodal ultrasound-clinical model, multimodal ultrasound-radiomics model and combined model were constructed using logistic regression algorithm. The receiver operating characteristic curves were drawn to evaluate the efficacy of these models. Results The sensitivity, specificity, accuracy and the area under the curve (AUC) of multimodal ultrasound-clinical model, multimodal ultrasound-radiomics model and combined model in training set was 79.83%, 76.51%, 78.91% and 0.826, 73.41%, 75.01%, 69.20% and 0.775, 87.20%, 88.19%, 84.56% and 0.876, respectively, which was 43.82%, 100%, 60.91% and 0.750, 88.24%, 57.14%, 79.17% and 0.741, 87.02%, 85.48%, 85.70% and 0.875 in testing set, respectively. Conclusion Combined radiomics and multimodal ultrasound had good efficacy for differentiating benign and malignant superficial lymph nodes. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
