摘要
目的探讨基于治疗前MRI-T2WI影像组学特征预测宫颈鳞癌中、低分化的可行性。方法回顾性分析72例宫颈癌患者治疗前MRI图像资料,其中中分化患者40例,低分化患者32例,运用IBEX工具箱提取患者肿瘤大体肿瘤靶区的影像组学特征,特征筛选采用t检验,LASSO算法用于特征降维,Pearson相关用于分析组学特征与病理类型的相关性,特征参数绘制接受者操作特性(Receiver Operating Characteristic,ROC)曲线,并计算ROC曲线下面积(Area Under Curve,AUC)。通过ROC分析各个坐标的最大YOUDEN指数并计算组学特征的敏感性、特异性和最佳阈值。结果提取的682个组学特征中,有8个特征与宫颈鳞癌中、低分化病理学分级具有相关性,相关系数均大于0.4,AUC值均大于0.7。其中333_4相关和2_1相关特征参数分别在中分化和低分化病理类型中表现出最佳预测性能,AUC值分别为0.808、0.828,敏感性和特异性分别为0.667、0.663和0.867、0.917。结论MRI-T_(2)WI图像的影像组学特征可以作为治疗前宫颈鳞癌中、低分化病理学分级诊断的一种无创、有价值的辅助手段。
Objective To explore the feasibility of predicting moderate and low differentiation of cervical squamous cell carcinoma based on MRI-T_(2)WI radiomics characteristics before treatment.Methods MRI images of 72 patients with cervical cancer before treatment were retrospectively analyzed,including 40 cases with moderate differentiation and 32 cases with low differentiation.IBEX toolbox was used to extract the radiomics features of the gross tumor volume of patients’tumor.The t test was used for feature screening,LASSO algorithm was used for feature dimension reduction,Pearson correlation was used to analyze the correlation between radiomics features and pathological types.The receiver operating characteristic curve was drawn and the area under the receiver operating characteristic(ROC)curve was calculated.ROC was used to analyze the maximum YOUDEN index of each coordinate and calculate the sensitivity,specificity and optimal threshold of the omics characteristics.Results A total of the 682 radiomics features extracted,8 features were correlated with moderate and low differentiation pathological grades of cervical squamous cell carcinoma,with correlation coefficients greater than 0.4 and AUC values higher than 0.7.Among them the 333_4 correlation and the 2_1 correlation showed the best predictive performance in moderate differentiated and low differentiation pathological types,with AUC values of 0.808 and 0.828,respectively.The sensitivity and specificity were 0.667,0.663 and 0.867,0.917,respectively.Conclusion The radiomics features of MRI-T2WI can be used as a non-invasive and effective adjunct to the pathological grade diagnosis of moderate and low differentiation of pretherapeutic heterogeneity of cervical squamous cell carcinoma.
作者
党俊明
朱超华
黄慧娴
陆合明
DANG Junming;ZHU Chaohua;HUANG Huixian;LU Heming(Ruikang Clinical Medical College,Guangxi University of Chinese Medicine,Nanning Guangxi 530000,China;Department of Radiation Oncology,People’s Hospital of Guangxi Zhuang Autonomous Region,Nanning Guangxi 530000,China)
出处
《中国医疗设备》
2022年第11期71-75,84,共6页
China Medical Devices
基金
广西自然科学基金(2020GXNSFBA297091)。
关键词
宫颈鳞癌
影像组学
病理分型
特征降维
肿瘤异质性
cervical squamous cell carcinoma
radiomics
pathological type
feature dimension
tumor heterogeneity
