磁共振影像组学在预测局部晚期宫颈鳞癌患者病理分型、临床分期及行根治性同步放化疗后评价疗效的价值分析  

Value of magnetic resonance imaging omics in predicting pathological classification, clinical staging, and evaluating efficacy after radical synchronous radiotherapy and chemotherapy in locally advanced cervical squamous cell carcinoma patients

作　　者：杨凤玲 林灿阳 廖东霞 倪晓雷[1] YANG Feng-ling;LIN Can-yang;LIAO Dong-xia(Department of Comprehensive Tumor Therapy,Longyan First Hospital of Fujian Province,Longyan Fujian 364000,China)

机构地区：[1]龙岩市第一医院肿瘤综合治疗科,福建龙岩364000

出　　处：《临床和实验医学杂志》2023年第22期2430-2435,共6页Journal of Clinical and Experimental Medicine

基　　金：福建省自然科学基金资助项目(编号:2020J011325)。

摘　　要：目的探讨基于磁共振影像组学在预测局部晚期宫颈鳞癌患者病理分型、临床分期及行根治性同步放化疗后评价疗效的价值。方法回顾性选择2017年1月至2020年12月来龙岩市第一医院诊治的120例宫颈癌患者作为研究对象。最终116例宫颈癌患者纳入研究,根据3∶1比例随机分为训练集(n=87)、测试集(n=29)。所有患者在治疗后第2周末、治疗后第4周末行盆腔的平扫MRI、动态增强MRI、DWI,根据支持向量机、逻辑回归(LR)、随机森林3种分类器方法,分别建立磁共振影像组学模型。分析局部晚期宫颈鳞癌患者的临床基线资料,根据T1-micro确定不同分类器模型性能,分析3种分类器在DWI、T_(2)WI、增强T_(1)WI序列对局部晚期宫颈鳞癌患者的病理分型、临床分期、疗效的预测价值。结果训练集和测试集患者的年龄、临床分期、月经情况、病理类型、临床分期、疗效、最大径比较,差异均无统计学意义(P>0.05)。最终选择LR作为最佳的分类器。对宫颈鳞癌角化型与非角化型、临床分期诊断中,不管在训练集还是测试集中,3个序列影像组学模型中的曲线下面积(AUC)比较,差异无统计学意义(P>0.05),然而多序列联合模型的AUC值明显较单一序列模型高,差异有统计学意义(P<0.05);对宫颈鳞癌根治性同步放化疗疗效评价时,在训练集中多序列联合模型的AUC值明显较单一序列模型高,差异有统计学意义(P<0.05)。在测试集中,多序列联合模型与T_(2)WI相比,AUC明显较高,然而与DWI、增强T_(1)WI序列模型中AUC比较,差异无统计学意义(P>0.05)。结论基于磁共振影像组学构建多序列联合模型可用于预测局部晚期宫颈鳞癌患者病理分型、临床分期及根治性同步放化疗后疗效评价。Objective To explore the value of magnetic resonance imaging omics in predicting pathological classification,clinical staging,and evaluating the efficacy of radical synchronous radiotherapy and chemotherapy in locally advanced cervical squamous cell carcinoma patients.Methods A total of 120 patients with cervical cancer diagnosed and treated in Longyan First Hospital of Fujian Province from January 2017 to December 2020 were retrospectively selected as the study subjects.Finally,116 cervical cancer patients were included in the study and randomly divided into a training set(n=87)and a testing set(n=29)based on a 3∶1 ratio.All patients received plain scan MRI,dynamic enhanced MRI and DWI of the pelvic cavity at the 2nd week of treatment and the 4th weekend after treatment.According to three classifier methods,namely support vector machine,logical regression(LR)and random forest,the MRI model was established respectively.The clinical baseline data of locally advanced cervical squamous cell carcinoma patients was analyze,the performance of different classifier models based on T1-micro was determine,and the predictive value of three classifiers on DWI,T_(2)WI,and enhanced T_(1)WI sequences for pathological classification,clinical staging,and efficacy of locally advanced cervical squamous cell carcinoma patients was analyze.Results There was no statistically significant difference in age,clinical stage,menstrual status,pathological type,clinical stage,efficacy,and maximum diameter between the training and testing sets of patients(P>0.05).Finally,LR was chosen as the optimal classifier.In the diagnosis of keratinized and non-keratinized cervical squamous cell carcinoma,as well as clinical staging,there was no statistically significant difference in area under the curve(AUC)between the three sequence imaging omics models in both the training and testing sets(P>0.05).However,the AUC value of the multi sequence combined model was significantly higher than that of the single sequence model,the difference was statisticall

关 键 词：磁共振影像组学 局部晚期宫颈鳞癌 病理分型 临床分期 根治性同步放化疗 疗效评价 

分 类 号：R737.33[医药卫生—肿瘤] R445.2[医药卫生—临床医学]