基于支持向量机的全乳纹理分析预测乳腺癌新辅助化疗疗效的价值  

Prediction of Neoadjuvant Chemotherapy Outcomes of Breast Cancer Based on Whole Breast Texture with Support Vector Machine

作　　者：周靖宇[1] 单慧明[1] 成官迅[1] ZHOU Jing-yu;SHAN Hui-ming;CHENG Guan-xun(Department of Radiology,Peking University Shenzhen Hospital,Shenzhen 518000,Guangdong,China)

机构地区：[1]北京大学深圳医院医学影像科,广东深圳518000

出　　处：《医学信息》2022年第19期60-64,共5页Journal of Medical Information

基　　金：北京大学深圳医院科研项目(编号:LCYJ2017006)。

摘　　要：目的探讨基于支持向量机的全乳纹理分析预测局部进展晚期乳腺癌新辅助化疗疗效(NAC)的应用价值。方法纳入2013年9月-2019年9月北京大学深圳医院收治的乳腺癌患者152例,均行3次动态增强磁共振检查(治疗前标记为E1、NAC 2个周期后标记为E2、NAC 4个周期后标记为E3),完成8个周期标准NAC后接受根治性乳房切除。根据病理Miller-Payne评级分为组织学显著反应组(MHR)与组织学非显著反应组(NMHR)。从动态增强磁共振图像中提取基于灰度直方的均值、方差、一致性、偏度、峰度以及灰度共生矩阵的能量、熵、惯性矩、相关性、逆差距的患侧全乳纹理特征,在R project构建径向基函数内核的支持向量机模型预测乳腺癌NAC疗效,其中训练病例114例,测试病例38例,分析模型预测准确度及受试者工作特征曲线、曲线下面积。结果152例病例中MHR组54例,NMHR组98例。两组不同年龄、月经状态、家族史、肿瘤分期、分子分型及化疗方案比较,差异无统计学意义(P>0.05);NMHR组逆差距(E1、E2)、灰度一致性(E2、E3)及比值(E3/E1)、灰度偏度比值(E2/E1、E3/E1)和峰度比值(E3/E1)高于MHR组,而惯性矩(E2)低于MHR组,差异有统计学意义(P<0.05);逆差距(E1、E2)、惯性矩(E2)、灰度一致性(E2、E3)及比值(E3/E1)、灰度偏度比值(E2/E1、E3/E1)和峰度比值(E3/E1)单独预测NAC疗效的AUC值介于0.5797~0.6419,诊断效能较低。E1+E2、E1+E3、E1+E2+E3三种组合模型中,E1+E2+E3组合的预测效能最高,在训练组准确度为77.20%,曲线下面积为0.90,在测试组准确度为71.10%,曲线下面积为0.76。结论基于支持向量机的全乳纹理可以预测乳腺癌新辅助化疗疗效。Objective To explore the application value of whole breast texture analysis based on support vector machine in predicting the efficacy of neoadjuvant chemotherapy(NAC)in locally advanced breast cancer.Methods A total of 152 patients with breast cancer admitted to Peking University Shenzhen Hospital from September 2013 to September 2019 were enrolled.All patients underwent three dynamic contrast-enhanced magnetic resonance examinations(before treatment marked as E1,after 2 cycles of NAC marked as E2,after 4 cycles of NAC marked as E3),and underwent radical mastectomy after 8 cycles of standard NAC.According to the pathological Miller-Payne rating,the patients were divided into major histological response group and non-major histological response group.The mean,variance,consistency,skewness,kurtosis based on the gray square and the energy,entropy,moment of inertia,correlation and inverse gap of the gray level co-occurrence matrix were extracted from the dynamic enhanced magnetic resonance image.The texture features of the affected side of the whole breast were extracted.The support vector machine model with radial basis function kernel was constructed in R project to predict the NAC efficacy of breast cancer,including 114 training cases and 38 test cases.The prediction accuracy of the model and the receiver operating characteristic curve and area under the curve were analyzed.Results Among 152 patients,there were 54 cases of MHR group and 98 cases of NMHR group.There was no significant difference in age,menstrual status,family history,tumor stage,molecular classification and chemotherapy regimen between the two groups(P>0.05).The inverse difference(E1,E2),gray consistency(E2,E3)and ratio(E3/E1),gray skewness ratio(E2/E1,E3/E1)and kurtosis ratio(E3/E1)in the NMHR group were higher than those in the MHR group,while the moment of inertia(E2)was lower than that in the MHR group,the difference was statistically significant(P<0.05).The AUC values of inverse difference(E1,E2),moment of inertia(E2),gray consistency(E2,E3)a

关 键 词：乳腺癌 新辅助化疗 纹理特征 支持向量机 

分 类 号：R737.9[医药卫生—肿瘤]