摘要
目的:探讨超声瘤内+瘤周影像组学模型预测乳腺癌雌激素受体和孕激素受体双阴性表达的价值。方法:回顾性分析南京医科大学第三附属医院(常州市第二人民医院)2019年1月至2021年3月间符合条件的374例乳腺癌患者,按照7:3的比例随机划分为训练集(n=261)和测试集(n=113)。使用Pyradiomics包从每例患者超声图像的瘤内、瘤周5 mm感兴趣区域中提取影像组学特征,采用独立样本t检验、Spearman相关性分析和最小绝对收缩和选择算子筛选最佳影像组学特征,并使用支持向量机分类器分别构建瘤内、瘤周、瘤内+瘤周影像组学模型。受试者操作特征曲线下面积(AUC)、校准曲线和决策曲线被用来评价模型性能和实用性。结果:分别筛选出12个瘤内、11个瘤周、17个瘤内+瘤周影像组学特征构建预测模型。其中,瘤内+瘤周影像组学模型预测效果最佳,在训练集和测试集AUC分别为0.890、0.848,优于单一瘤内或瘤周影像组学模型。校准曲线和决策曲线分析也表明瘤内+瘤周影像组学模型具有良好的校准度和临床实用性。结论:超声瘤内+瘤周影像组学是无创预测乳腺癌雌激素受体和孕激素受体双阴性表达的有效手段。
Objective To investigate the value of ultrasound intratumoral+peritumoral radiomics model in predicting doublenegative expression of estrogen receptor and progesterone receptor in breast cancer.Methods A retrospective analysis was conducted on 374 breast cancer patients admitted to the Third Affiliated Hospital of Nanjing Medical University(the Second People's Hospital of Changzhou)between January 2019 and March 2021.The patients were randomly divided into a training set(n=261)and a test set(n=113)at a ratio of 7:3.Pyradiomics package was used to extract radiomics features from the intratumoral and 5 mm peritumoral regions of interest in ultrasound images.After selecting the optimal radiomics features using the independent samples t-test,Spearman's correlation analysis,and least absolute shrinkage and selection operator,support vector machine classifier was employed to construct intratumoral,peritumoral,and intratumoral+peritumoral radiomics models,separately.Area under the receiver operating characteristic curve(AUC),calibration curve,and decision curve were used to evaluate the model performance and utility.Results Twelve intratumoral,11 peritumoral,and 17 intratumoral+peritumoral radiomics features were screened to construct the corresponding predictive models.Among the 3 models,the intratumoral+peritumoral radiomics model had the best predictive performance,with AUC values of 0.890 in the training set and 0.848 in the test set,outperforming single intratumoral or peritumoral radiomics model.Calibration curve and decision curve also demonstrated its high calibration and clinical utility.Conclusion Ultrasound radiomics combining intratumoral and peritumoral features serves as an effective non-invasive approach for predicting estrogen and progesterone receptor double-negative expression in breast cancer.
作者
马晓峰
张恒
倪昕晔
MA Xiaofeng;ZHANG Heng;NI Xinye(Department of Radiology,Jiangyin No.3 People's Hospital,Jiangyin 214400,China;Department of Radiotherapy,the Third Affiliated Hospital of Nanjing Medical University(the Second People's Hospital of Changzhou)/Jiangsu Province Engineering Research Center of Medical Physics/Center for Medical Physics,Nanjing Medical University/Key Laboratory of Medical Physics in Changzhou,Changzhou 213003,China)
出处
《中国医学物理学杂志》
2025年第12期1585-1592,共8页
Chinese Journal of Medical Physics
基金
国家自然科学基金(62371243)
江苏省医学重点学科建设单位(肿瘤治疗学(放疗)JSDW202237)
江苏省重点研发计划社会发展项目(BE2022720)
江苏省卫健委面上项目(M2020006)
江苏省自然科学基金(BK20231190)
常州市社会发展项目(CE20235063)。
关键词
乳腺癌
雌激素受体
孕激素受体
超声影像组学
breast cancer
estrogen receptor
progesterone receptor
ultrasound radiomics
