基于高分辨率T2WI的影像组学对直肠癌EGFR表达的预测价值  

High-resolution T2-weighted image-based radiomics in the preoperative EGFR expression of rectal cancer

作　　者：胡友强 罗敏[1] 董小娟 邹龙权 谢刚[1] 张远林 刘文军[1] 姜萍[1] HU You-qiang;LUO-Min;DONG Xiao-juan;ZOU Long-quan;XIE Gang;ZHANG Yuan-lin;LIU Wen-jun;JIANG Ping(Department of Radiology,Zigong Fourth People's Hospital,Zigong 643000;Department of Pathology,Zigong Fourth People's Hospital,Zigong 643000;Department of Radiology,Zigong Third People's Hospital,Zigong 643020,Sichuan,China)

机构地区：[1]自贡市第四人民医院放射科,四川自贡643000 [2]自贡市第四人民医院病理科,四川自贡643000 [3]自贡市第三人民医院放射科,四川自贡643020

出　　处：《川北医学院学报》2024年第3期358-362,共5页Journal of North Sichuan Medical College

基　　金：四川省自贡市重点研发计划项目(2020YLSF17)。

摘　　要：目的:探讨基于高分辨率T2WI的影像组学对直肠癌EGFR表达状态的预测价值。方法:回顾性分析经术后病理确诊且在接受治疗前行MRI检查的208例直肠癌患者的临床及影像资料,根据EGFR表达水平不同将患者分为阳性组和阴性组。在高分辨率T2WI图像上勾画病灶的三维容积兴趣区(VOI)并提取影像组学特征,将208例患者分为训练集(n=145)和测试集(n=63),并对特征进行降维,将降维后的特征建立支持向量机(SVM)、逻辑回归(LR)、随机森林(RF)及线性判别分析(LDA)四种分类器学习模型,分别绘制训练集和测试集的受试者工作特征(ROC)曲线,并获得曲线下面积(AUC)。结果:208例患者中,EGFR阳性表达99例(47.6%)。二元Logistic回归分析显示示低分化和淋巴结转移是EGFR阳性表达的独立危险因素(P<0.05)。训练集与测试集的患者在性别、年龄、TN分期及分化程度差异均无统计学意义(P>0.05)。4种影像组学模型均有一定的预测效能,其中SVM模型训练集与测试集的诊断效能均为最高,在训练集和测试集中的AUC分别为0.803、0.725。结论:基于高分辨率T2WI图像构建的影像组学模型对直肠癌EGFR表达状态具有一定预测价值。Objective:To investigate the value of high-resolution T2-weighted(T2WI)image-based radiomics in the preoperative EGFR expression of rectal cancer.Methods:This was a retrospective analysis of 208 patients with a pathology confirmed rectal cancer diagnosis who underwent surgery and high-resolution T2WI magnetic resonance imaging(MRI)before surgery.Patients were divided into two groups:high EGFR expression and low EGFR expression.The volume of interest(VOI)were drawn based on high-resolution T2WI images and then radiomics features were extracted.The samples were divided into a training set(n=145)and a test set(n=63)for machine learning.Four machine learning classifier models were established:support vector machine(SVM),logistic regression(LR),random forests(RF),and linear discriminant analysis(LDA).Receiver operating characteristic(ROC)curves and the area under the curve(AUC)of the training set and test set were obtained.Results:A total of 208 patients were enrolled,including 99 patients with high EGFR expression(47.6%).Logistic regression analysis showed that low differentiation and lymph node metastasis were independent risk factors of EGFR expression(P<0.05).There was no significant difference of gender,age,TN stage and differentiation degree between training set and testing set(P>0.05).All the four radiomics models had certain diagnostic efficacy.SVM algorithm had the highest diagnostic accuracy,the AUCs in the training and testing groups were 0.803 and 0.725,respectively.Conclusion:The high-resolution T2WI image-based radiomics could be a valuable preoperative test for predicting EGFR expression of rectal cancer.

关 键 词：直肠肿瘤 影像组学 磁共振成像 EGFR 

分 类 号：R445[医药卫生—影像医学与核医学]