能谱CT影像组学预测结直肠癌MSI状态的应用研究  

The Research of Application of Predicting MSI in Colorectal Cancer Based on Spectral CT Radiomics

作　　者：张红霞 田为中 夏建国[1] 杨维柘 ZHANG Hong-xia;TIAN Wei-zhong;XIA Jian-guo;YANG Wei-zhe(The Affiliated Taizhou People's Hospital of Nanjing Medical University,Taizhou Jiangsu 225300,China)

机构地区：[1]南京医科大学附属泰州人民医院,江苏泰州225300

出　　处：《泰州职业技术学院学报》2024年第4期85-89,共5页Journal of Taizhou Polytechnic College

基　　金：江苏大学2021年度临床医学科技发展基金项目(JLY2021183,项目主持人:杨维柘);2023年度泰州市科技支撑计划项目(SSF20230116,项目主持人:田为中).

摘　　要：目的研究基于能谱CT成像静脉期单能量序列影像组学模型,用于预测结直肠癌微卫星不稳定(MSI)状态。方法回顾性收集2019年7月至2022年8月在泰州市人民医院经手术切除或穿刺活检病理确诊为结直肠癌患者97例(男性57例、女性40例)。根据免疫组化MSI表达结果进行分组:MSI组34例、微卫星稳定(MSS)组63例。以DICOM格式将静脉期1.25mm原始图像数据传至GEAW4.7后处理工作站,生成碘基物质分解图、70keV、100keV单能量图,导入ITK-SNAP开源软件,手动勾画感兴趣区(ROI),利用Pyradiomics工具进行影像组学特征提取,采用mRMR(最大相关和最小冗余)和LASSO(最小绝对收缩选择算子)算法对训练组的影像组学特征进行降维,获取关键的影像组学特征,建立预测肿瘤生物学行为的影像组学模型。采用受试者工作特征(ROC)曲线下面积(AUC)来评估模型的诊断效能。使用校正曲线对模型的校正性能进行评估,应用决策曲线评价模型在训练组中的临床实用性。结果静脉期单能量70keV序列、100keV序列、碘基物质分解图共三种模型各提取了1218个特征,其中70keV单能量序列特征分类性能最佳。结论基于能谱CT成像的单能量模型可以有效地区分结直肠癌MSI和MSS状态患者,可以作为术前预测结直肠癌患者MSI状态的重要影像生物标志物,为指导临床医生制定个体化治疗方案及评估患者预后提供新思路。Objective To study the radiomics model based on single energy sequence of spectral CT imaging in ve-nous phase for predicting microsatellite instability(MSI)status of colorectal cancer.Methods From July 2019 to August 2022,97 patients(57 males and 40 females)with colorectal cancer diagnosed by surgical resection or biop-sy in Taizhou People’s Hospital were retrospectively collected.According to the results of immunohistochemical MSI expression,the patients were divided into MSI group(34 cases)and MSS group(63 cases).The raw image data of 1.25 mm venous images were transmitted to GE AW4.7 post-processing workstation in DICOM format to generate iodine-based material decomposition maps,70keV,100keV monoenergetic maps.The images were imported into ITK-SNAP open source software,and the region of interest(ROI)was manually delineated.The radiomics feature extraction was performed using the Pyradiomics tool.The mRMR(maximum relevance and minimum redundancy)and LASSO(least absolute shrinkage and selection operator)algorithms were used to re-duce the dimension of the radiomics features of the training group and obtain the key radiomics features,and the radiomics model for predicting tumor biological behavior was established.The area under the ROC curve(AUC)analysis was used to evaluate the diagnostic efficacy of the ten models.The calibration curve was used to evaluate the calibration performance of the model,and the decision curve was used to evaluate the clinical practi-cability of the model in the training group.Results Venous phase single energy 70keV,100keV sequence,io-dine-based material decomposition diagram,sequence diagram extracted 1218 features respectively,among which the venous phase 70 keV single energy sequence had the best classification performance.Conclusion The single-energy model based on energy spectrum CT imaging can distinguish MSI status from MSS status effective-ly,It can be used as an important imaging biomarker to predict the status of MSI in colorectal cancer patients be-fore operation,and pro

关 键 词：能谱CT成像 影像组学 结直肠癌 微卫星不稳定性 

分 类 号：R814.42[医药卫生—影像医学与核医学]