基于瘤内及瘤周超声组学评分联合临床因素的列线图模型可较好预测乳腺癌新辅助化疗疗效  

The nomogram model based on the combination of intratumoral and peritumoral ultrasound radiomics scores and clinical factors demonstrates good predictive value for the efficacy of neoadjuvant chemotherapy in breast cancer

作　　者：许雨晴 杨茹 慈红非 彭潇雨 李阳 XU Yuqing;YANG Ru;CI Hongfei;PENG Xiaoyu;LI Yang(Department of Ultrasound,The First Affiliated Hospital of Bengbu Medical University,Bengbu 233000,China;School of Graduate,Bengbu Medical University,Bengbu 233000,China;Department of Pathology,The First Affiliated Hospital of Bengbu Medical University,Bengbu 233000,China)

机构地区：[1]蚌埠医科大学第一附属医院超声科,安徽蚌埠233000 [2]蚌埠医科大学研究生院,安徽蚌埠233000 [3]蚌埠医科大学第一附属医院病理科,安徽蚌埠233000

出　　处：《分子影像学杂志》2025年第3期264-271,共8页Journal of Molecular Imaging

基　　金：蚌埠市科技创新指导类项目(20230129);安徽省卫生健康科研项目(AHWJ2023A20372)。

摘　　要：目的 探讨基于乳腺癌治疗前超声图像的瘤内及瘤周超声组学特征联合临床因素构建联合模型来预测乳腺癌新辅助化疗的疗效。方法 回顾性分析于我院接受新辅助化疗治疗的女性乳腺癌患者145例,按照术后病理结果分为病理完全缓解组(n=85)和非病理完全缓解组(n=60)。使用3D Slicer软件勾画瘤内及瘤周区域并提取影像组学特征,构建影像组学模型并通过支持向量机算法获得影像组学评分,筛选临床特征构建临床模型。选择效能最高组学模型的影像组学评分和临床独立预测因素联合构建列线图模型,绘制ROC曲线评估各模型预测效能。结果 与单个瘤内及瘤周模型相比,瘤内+瘤周模型的预测效果较好。筛选出雌激素受体、人表皮生长因子受体-2为独立预测因素构建临床模型,训练集和验证集的曲线下面积分别为0.707、0.778。列线图模型较于各模型诊断性能均有提高,训练集和验证集曲线下面积提高至0.874、0.885。结论 基于治疗前超声图像的乳腺癌原发灶瘤内+瘤周影像组学评分联合临床因素构建的列线图模型对乳腺癌新辅助化疗疗效具有较好的预测价值,有望指导临床决策。Objective To explore an integrated model that combines intratumoral and peritumoral sonographic features along with clinicopathological factors for predicting the response to neoadjuvant chemotherapy in breast cancer.Methods A retrospective analysis was conducted on a total of 145 female breast cancer patients who underwent neoadjuvant chemotherapy in our hospital.Based on postoperative pathological results,the patients were divided into a pathological complete response(pCR)group(n=85)and a non-pCR group(n=60).The intratumoral and peritumoral regions were demarcated,and radiomics features were extracted using 3D Slicer software.The radiomics model was constructed,and radiomics scores were obtained through the support vector machine algorithm.The radiomics score with the highest performance from the radiomics model and the clinical independent predictors were selected to construct a nomogram model.The ROC curve was drawn to assess the predictive performance of each model.Results In comparison to single intratumoral and peritumoral models,the intratumoral+peritumoral model exhibited a superior prediction effect.ER and HER-2 were chosen as independent predictors to construct a clinical model.The AUC of the training set and validation set were 0.707 and 0.778 respectively.The diagnostic performance of the nomogram model was enhanced compared to the other models,and the AUC of the training set and validation set was increased to 0.874 and 0.885.Conclusion The nomogram model based on the intratumoral and peritumoral radiomics scores of primary breast cancer derived from pre-treatment ultrasound images in combination with clinical factors holds significant predictive value for the efficacy of neoadjuvant chemotherapy in breast cancer.This model is expected to guide clinical decision-making.

关 键 词：乳腺癌 新辅助治疗 影像组学 瘤周 列线图 

分 类 号：R73[医药卫生—肿瘤]