摘要
目的探讨瘤内及不同范围瘤周影像组学对行同步放化疗(concurrent chemoradiothrapy,CCRT)局部中晚期宫颈癌患者无进展生存期(progression-free survival,PFS)的预测价值。方法回顾性选取135例经病理证实的宫颈癌患者,包括32例进展和103例无进展患者,以7∶3的比例分为训练集和验证集。基于动态对比增强磁共振成像(dynanic contrast-enhanced magnetic resonance imaging,DCE-MRI)第二期图像在瘤内和3、5、7 mm瘤周区域进行三维容积感兴趣区(volume of interest,VOI)勾画,分别提取影像组学特征并降维,将筛选出的特征构建瘤内、瘤周、瘤内联合瘤周影像组学模型,比较预测效能。保留有统计学意义的临床特征构建临床模型。基于AUC最佳影像组学特征与筛选出的临床特征共同建立综合模型。利用AUC及一致性指数(consistency index,C-index)评估模型预测能力。AUC及C-index值最高的模型继续进行校准曲线、决策曲线分析(decision curve analysis,DCA)及Kaplan-Meier生存曲线后续评估。结果瘤内+5 mm瘤周模型较其他范围影像组学模型显示出较好的预测效能,AUC为0.852。相较于临床和影像组学模型,综合模型显示出最优的预测效能,AUC分别为0.766、0.852、0.872。经校准曲线和DCA分析,综合模型校准度较高,临床净收益较大,Kaplan-Meier生存曲线可区分出发生疾病进展的高风险患者和低风险患者。结论基于DCE-MRI的瘤内联合瘤周影像组学特征可作为评估行CCRT局部中晚期宫颈癌患者PFS的有效指标,其中瘤内+5 mm瘤周模型显示出较高的预测能力,且纳入临床参数的综合模型效能更佳。
Objective To explore the predictive value of intratumoral and peritumoral radiomics in different ranges for progression-free survival(PFS)in patients with locally advanced cervical cancer undergoing concurrent chemotherapy(CCRT).Methods A total of 135 patients with cervical cancer were retrospectively selected,including 32 patients with progression and 103 patients without progression.They were divided into the training set and the validation set in a 7∶3 ratio.On the basis of the second phase images of dynanic contrast-enhanced magnetic resonance imaging(DCE-MRI),three-dimensional volume of interest(VOI)delimitations were performed in the 3,5,and 7 mm areas within and around the tumor.Radiomics features were extracted and dimensionally reduced,respectively.The selected charac-teristics were used to construct combined intratumoral,peritumoral,and intratumoral-peritumoral radiomic models to compare predictive efficacy.Clinical models were constructed by retaining statistically significant clinical characteristics.A comprehensive model was jointly established based on the best radiomic features of area under curve(AUC)and the screened clinical features.The predictive ability of the model was evaluated using AUC and the consistency index(C-index).The models with the highest AUC and C-index values were used to evaluate the calibration curve,decision curve analysis(DCA)and Kaplan-Meier survival curve.Results The intratumoral+5 mm peritumoral model showed better predictive efficacy than other radiomic models,with an AUC of 0.852,Compared to the clinical and radiomic models,the comprehensive model showed the best predictive efficacy,with AUCs of 0.766,0.852,and 0.872,respec-tively.Through the calibration curve and DCA analysis,the comprehensive model had a high degree of calibration and a large clinical net benefit.The Kaplan-Meier survival curve could distinguish between high-risk patients and low-risk patients with disease progression.Conclusion The combined intratumoral and peritumoral radiomic characteristics based on DCE-MRI can be used as an effective indicator for evaluating PFS in patients with locally advanced cervical cancer undergoing CCRT.Among them,the intratumoral+5 mm peritumoral model shows higher predictive ability,and the comprehensive model incorporating clinical parameters has better efficacy.
作者
王磊
常霄
王梓萌
李娇娇
崔书君
杨飞
朱月香
WANG Lei;CHANG Xiao;WANG Zimeng;LI Jiaojao;CUI Shujun;YANG Fei;ZHU Yuexiang(Graduate School,Hebei North University,Zhangjiakou 075000,Hebei,China;Department of Medical Imaging,The First Affiliated Hospital of Hebei North University,Zhangjiakou 075061,Hebei,China)
出处
《山东大学学报(医学版)》
北大核心
2025年第6期45-54,共10页
Journal of Shandong University:Health Sciences
基金
河北省自然科学基金(H2023405031)
河北省资助临床人才培养项目[冀卫办科教(2021)9号]。
关键词
宫颈癌
核磁共振
影像组学
瘤周微环境
生存预后
Cervical cancer
Nuclear magnetic resonance
Imaging omics
Peritumor microenvironment
Survival prognosis
