基于DCE-MRI影像组学定量预测进展期宫颈鳞癌同步放化疗反应的价值  

Value of Quantitative Prediction of Response to Concurrent Chemotherapy and Radiation Therapy in Advanced Cervical Squamous Cell Carcinoma Based on DCE-MRI Radiomics

作　　者：苏亚英 石子馨 张苗 焦光丽 杨飞[2] 崔书君[2] SU Ya-ying;SHI Zi-xin;ZHANG Miao;JIAO Guang-li;YANG Fei;CUI Shu-jun(Graduate School,Hebei North University,Zhangjiakou,Hebei 075000,China;Department of Radiology,The First Hospital Affiliated to Hebei North University,Zhangjiakou,Hebei 075000,China)

机构地区：[1]河北北方学院研究生院,河北张家口075000 [2]河北北方学院附属第一医院影像科,河北张家口075000

出　　处：《河北北方学院学报（自然科学版）》2023年第2期11-17,共7页Journal of Hebei North University:Natural Science Edition

基　　金：河北省政府资助临床医学人才培养项目;河北省医学科学研究课题计划项目(No.20210597)。

摘　　要：目的探讨基于动态对比增强磁共振成像(dynamic contrast-enhanced magnetic resonance imaging,DCE-MRI)的肿瘤内及含肿瘤外缘5 mm移行带的瘤周影像组学标签对进展期宫颈鳞癌同步放化疗(concurrent chemotherapy and radiation therapy,CCRT)反应的预测价值。方法回顾性收集126例经病理活检证实的宫颈鳞癌患者DCE-MRI图像及临床资料,按照7∶3的比例随机分为训练组(n=88)和验证组(n=38)。使用3D-Slicer软件在两个靶区(VOI、VOI+5mm)提取影像组学特征,采用最大相关最小冗余(mRMR)及最小绝对收缩与选择算子(LASSO)算法进行特征筛选并构建影像组学标签。利用ROC曲线及决策曲线(DCA)分析评估影像组学标签的预测效能及临床应用价值。结果VOI+5mm瘤周影像组学标签在训练组和验证组中的AUC(分别为0.855和0.819)均高于VOI瘤内影像组学标签(分别为0.770和0.750),瘤内标签和瘤周标签在训练组中预测CCRT反应的敏感度和特异度分别为74.5%和70.7%、76.6%和82.9%,而在验证组中分别为60.0%和83.3%、80.0%和77.8%。DCA显示当阈值概率分别为0.06~0.71和0.01~1.00时,瘤内标签和瘤周标签的临床净收益最大。结论基于DCE-MRI的影像组学标签在预测进展期宫颈鳞癌CCRT反应方面有一定临床应用价值,瘤周影像组学特征对于提升预测效能具有很大潜力,可为宫颈癌患者优化诊疗策略及改善预后提供参考。Objective To explore the value in predicting the response to concurrent chemotherapy and radiation therapy(CCRT)of advanced cervical squamous cell carcinoma by comparing the dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)-based peritumoral radiomics signature of 5 mm transitional zone with intratumoral ones.Methods The DCE-MRI images and clinical data of 126 patients with cervical squamous cell carcinoma confirmed by pathological biopsy were collected retrospectively.They were randomized into training group(n=88)and verification group(n=38)according to the proportion of 7:3.3D-Slicer software were used to extract radiomics features in two target volumes(VOI and VOI+5mm).The minimum redundancy-maximum relevance(mRMR)and the least absolute shrinkage and selection operator(LASSO)were applied for selecting features and constructing radiomics signatures.The predictive efficiency and clinical application value of the radiomics signatures were evaluated by receiver operating characteristic(ROC)curve and decision curve analysis(DCA).Results The area under the curve(AUC)of the VOI+5mmperitumoral radiomics signature(0.855 and 0.819)in the training and validation groups was higher than that in the VOI intratumoral radiomics signature(0.770 and 0.750).The sensitivity and specificity of the intratumoral and peritumoral signature for predicting response to CCRT in the training group were 74.5%and 70.7%,76.6%and 82.9%,respectively,while those in the validation group were 60.0%and 83.3%,80.0%and 77.8%,respectively.DCA showed that the clinical net benefit of the intratumoral and peritumoral signature reached the highest when the threshold probabilities were 0.06~0.71 and 0.01~1.00,respectively.Conclusions The radiomics signatures based on DCE-MRI has certain clinical value in predicting the response to CCRT in advanced cervical squamous cell carcinoma.The peritumor radiomics features have the potential to improve the predictive efficiency of the model,providing reference for optimizing diagnosis and treatment strate

关 键 词：磁共振成像 影像组学 宫颈癌 同步放化疗 瘤周微环境 

分 类 号：R445.2[医药卫生—影像医学与核医学] R711.74[医药卫生—诊断学] R737.33[医药卫生—临床医学]