CT增强影像组学模型对结直肠癌旁肿瘤沉积的诊断价值  

Diagnostic value of imaging omics models based on CT enhanced intratumoral and peritumoral imaging in adjacent tumor deposition of colorectal cancer

作　　者：刘燕[1] 罗锦文[1] 刘艳丽[1] 唐亚霞[1] LIU Yan;LUO Jinwen;LIU Yanli;TANG Yaxia(Department of Medical Imaging,the Fifth Affiliated Hospital of Guangzhou Medical University,Guangzhou 510700,China)

机构地区：[1]广州医科大学附属第五医院医学影像科,广东广州510700

出　　处：《分子影像学杂志》2025年第4期484-491,共8页Journal of Molecular Imaging

基　　金：广东省医学科研基金面上项目(A2023485);广州市教育局高校研究生科研项目(2024312248);广州市卫生健康科技一般引导项目(20241A011099)。

摘　　要：目的探讨基于CT增强瘤内和瘤周区域的影像组学模型在结直肠癌(CRC)旁肿瘤沉积(TDs)的诊断价值。方法回顾性分析2017年1月~2024年9月本院及TCIA数据库内经手术病理证实的330例CRC患者的CT增强图像,按照术后病理分为TDs阳性组(n=130)和TDs阴性组(n=200)。采用随机抽样的方法按照7∶3比例分为训练组(n=231)和测试组(n=99),在CT增强静脉期图像上手动逐层勾画感兴趣区(ROI),生成感兴趣体积,瘤周区域ROI分别按照2、4、6 mm等距外扩,使用Pyradiomics对各ROI区域进行特征提取,LASSO进行特征筛选,采用XGBoost的机器学习算法分别构建瘤内、瘤周及瘤内瘤周融合预测模型。采用ROC曲线来评价各模型的诊断性能,使用Delong检验来比较各模型预测性能。结果瘤内模型的训练组和测试组的ROC曲线下面积(AUC)分别为0.937、0.828。瘤周特征建模以瘤周4 mm范围诊断效能最佳,在训练组和测试组AUC分别为0.933、0.830。瘤内瘤周融合模型的预测诊断效能最佳,在训练组和测试组AUC分别为0.951、0.883,决策曲线分析显示瘤内瘤周融合模型对预测TDs净获益最佳。结论基于CT增强瘤内瘤周区域融合影像组学模型能很好预测CRC旁TDs,预测TDs净获益最佳,优于传统单一瘤内、瘤周影像组学模型,可辅助临床医师进行决策。Objective To explore the diagnostic value of a radiomics model based on the intratumoral and peritumoral regions in contrast-enhanced CT for peritumoral tumor deposits(TDs)in colorectal cancer(CRC).Methods A retrospective analysis was conducted on contrast-enhanced CT images of 330 CRC patients,confirmed by surgical pathology,from our hospital and the TCIA database between January 2017 and September 2024.Based on postoperative pathology,patients were classified into TDs-positive and TDs-negative groups.Using random sampling,patients were split into a training set(n=231)and a testing set(n=99)in a 7:3 ratio.Regions of interest(ROI)were manually delineated layer by layer on contrast-enhanced venous-phase images to generate volume of interest.The peritumoral ROIs were expanded outward by 2,4 and 6 mm.Radiomic features were extracted from each ROI using pyradiomics,and LASSO was employed for feature selection.XGBoost machine learning algorithm was used to construct separate prediction models for intratumoral,peritumoral,and combined intratumoral-peritumoral features.The diagnostic performance of each model was evaluated using ROC curves,and the DeLong test was used to compare the predictive performance of different models.Results The area under the ROC curve(AUC)for the intratumoral model was 0.937 in the training set and 0.828 in the testing set.Among the peritumoral models,the 4 mm peritumoral region exhibited the best diagnostic performance,achieving an AUC of 0.933 in the training set and 0.830 in the testing set.The combined intratumoral-peritumoral model demonstrated the highest predictive performance,with an AUC of 0.951 in the training set and 0.883 in the testing set.Decision curve analysis indicated that the combined model provided the highest net benefit for predicting TDs.Conclusion The radiomics model integrating intratumoral and peritumoral regions based on contrast-enhanced CT effectively predicts peritumoral TDs in CRC,offering the highest net benefit for TDs prediction.This model can assist clinician

关 键 词：计算机断层扫描 瘤内瘤周 影像组学 结直肠癌 肿瘤沉积 

分 类 号：R735.3[医药卫生—肿瘤]