摘要
目的:基于基线期ADC图全容积ROI特征提取构建影像组学预测模型,探讨其对肿块样乳腺癌新辅助化疗(NAC)疗效的预测价值。方法:将2016年1月-2017年12月在本院经粗针穿刺活检病理证实为浸润性乳腺癌的94例女性患者纳入研究,并按照7∶3的比例随机分为训练组(66例)和验证组(28例)。所有患者在NAC后采用Miller-Payne(M-P)病理分级系统评估疗效,其中获得病理完全缓解(PCR)者14例(14.9%)。回顾性分析每例患者NAC前、后的的临床资料及NAC前3.0T MRI检查资料。采用ITK-SNAP软件获得肿瘤的容积ROI,并采用A.K软件提取影像组学特征,序贯采用最大相关最小冗余(mRMR)特征选择方法和最小绝对收缩和选择算子(LASSO)进行降维分析,并构建影像组学模型(Radscore)。采用ROC曲线和临床决策曲线综合评价模型的预测效能和临床应用价值。结果:基于ADC图全容积ROI影像组学模型预测乳腺癌PCR,在训练组中的AUC和符合率分别为0.87(95%CI:0.75~0.99)和0.848,在验证组中分别为0.85(95%CI:0.71~1.00)和0.821。决策曲线显示,风险因素的概率阈值为10%~80%时,影像组学模型增加了更多的临床收益。结论:基于基线期ADC图全容积ROI影像组学模型对肿块样乳腺癌NAC后疗效(病理评估)有较好的预测效能。
Objective:To construct a radiomic prediction model based on pretreatment baseline apparent diffusion coefficient(ADC)mapping and whole volume ROI feature extraction,and to discuss its value in predicting the pathological complete response(PCR)to neoadjuvant chemotherapy(NAC)in mass-like breast cancer.Methods:94 female patients with invasive breast cancer confirmed by biopsy from January 2016 to December 2017 in the First Affiliated Hospital of Nanjing Medical University were included in this study,and were randomly divided into training group(66 cases)and validation group(28 cases)according to the ratio of 7:3.A total of 14 breast cancer patients was identified as PCR evaluated by Miller Payne(M-P)pathological grading system after NAC treatment.The clinical data of these patients before and after NAC and 3.0T MRI data before NAC were retrospective analyzed.The 3D region of interest(ROI)was manually segmented using ITK-SNAP software,and radiomics features were extracted using the A.K software.Maximum relevance minimum redundancy(mRMR)and least absolute shrinkage and selection operator(LASSO)were sequentially implied to reduce redundancy of radiomics features before radscore calculation.Receiver operation characteristic curve were used to validate the predictive efficacy of the model,then the decision curve was drawn to evaluate the clinical value of the model.Results:The performance of radiomics model to distinguish PCR from nPCR group was moderate.In the training set,the AUC and accuracy of the established radiomics model were 0.87(95%CI:0.75~0.99)and 0.848,respectively;in the validation set,the AUC and accuracy of radiomics model were 0.85(95%CI:0.71~1.00)and 0.821,respectively.Favorable clinical utility was observed under the risk threshold probability of 10%~80% using the decision curve analysis for the radiomics model.Conclusion:The radiomics model based on baseline ADC mapping and whole volume ROI can be used to predict the NAC efficacy of mass-like breast cancer.
作者
余雅丽
王晓
查小明
王思奇
娄鉴娟
邹启桂
王聪
蒋燕妮
YU Ya-Li;WANG Xiao;ZHA Xiao-ming(Department of Radiology,the First Affiliated Hospital of Nanjing Medical University,Nanjingc 210029,China)
出处
《放射学实践》
CSCD
北大核心
2022年第8期987-994,共8页
Radiologic Practice
基金
国家自然科学基金(81501442)。
关键词
乳腺癌
新辅助化疗
表观扩散系数
影像组学
全容积感兴趣区
磁共振成像
Breast cancer
Neoadjuvant chemotherapy
Apparent diffusion coefficient
Radiomics
Whole-volume interest region
Magnetic resonance imaging
