^(18)F-FDG PET/CT影像组学预测结外NK/T细胞淋巴瘤预后的价值  

Prognostic Value of Extranodal NK/T-Cell Lymphoma Based on^(18)F-FDG PET/CT Radiomics Features Combined with Clinical Factors

作　　者：罗与 魏巍[1] 余璇 白岩[1] 王梅云[1,3] LUO Yu;WEI Wei;YU Xuan;BAI Yan;WANG Meiyun(Department of Medical Imaging,Henan Provincial People’s Hospital,Zhengzhou 450003,China;Laboratory of Brain Science and Brain-Like Intelligence Technology,Institute for Integrated Medical Science and Engineering,Henan Academy of Sciences,Zhengzhou 450000,China)

机构地区：[1]河南省人民医院医学影像科,河南郑州450003 [2]郑州大学人民医院医学影像科,河南郑州450003 [3]河南省科学院医工融合研究所,脑科学与类脑智能实验室,河南郑州450000

出　　处：《中国医学影像学杂志》2024年第8期780-786,共7页Chinese Journal of Medical Imaging

基　　金：河南省医学科技攻关计划项目(SBGJ202101002)。

摘　　要：目的探讨^(18)F-FDG PET/CT影像组学模型对结外NK/T细胞淋巴瘤(ENKTL)无事件生存期的预测价值。资料与方法回顾性收集2013年1月—2021年1月于河南省人民医院治疗前行^(18)F-FDG PET/CT检查的ENKTL患者90例,随机分为训练组63例和验证组27例。从基线PET和CT图像上提取特征,应用最小绝对收缩和选择算子算法结合Cox生存分析筛选特征并构建临床模型、影像组学模型和临床+影像组学复合模型,以模型风险评分的中位数作为截断值将患者分为高危组和低危组。使用C指数和受试者工作特征曲线评价3个模型的预测性能。基于最优模型构建列线图,采用校准曲线描述最优模型预测ENKTL患者生存概率与实际概率的一致性,使用Kaplan-Meier分析和对数秩检验评估最优模型的预后价值。结果复合模型在训练组(C指数0.791,95%CI 0.702~0.879,曲线下面积为0.882)和验证组(C指数0.770,95%CI 0.650~0.889,曲线下面积为0.720)中比单独的临床模型和影像组学模型的预后表现更好。校准曲线表明复合模型在预测3年无事件生存期概率与实际结果的一致性较好,生存曲线显示高危组的无事件生存期显著低于低危组。结论基于影像组学和临床参数构建的复合模型可以提供更全面的预后信息并提高诊断准确性。列线图为ENKTL患者的风险分层提供了一种无创的诊断工具,有助于个体化治疗。Purpose To investigate the predictive value of^(18)F-FDG PET/CT radiomics model for event-free survival(EFS)in extranodal natural killer/T-cell lymphoma(ENKTL).Materials and Methods A total of 90 patients with ENKTL who underwent^(18)F-FDG PET/CT examination before treatment in Henan Provincial People′s Hospital from January 2013 to January 2021 were retrospectively collected and randomly divided into training(63 cases)and validation groups(27 cases).Features were extracted from baseline PET and CT images.The least absolute shrinkage and selection operator algorithm combined with Cox survival analysis were used to select features and construct clinical model,radiomics model and clinical+radiomics composite model,the median risk score of the model was used as the cut-off value to divide the patients into high-risk group and low-risk group.C-index and receiver operating characteristic curve analysis were used to evaluate the predictive performance of the three models.The nomogram was constructed based on the optimal model and calibration curves were used to describe the consistency between the survival probability and the actual probability of the optimal model for predicting ENKTL patients,Kaplan-Meier analysis and log-rank test were used to evaluate the prognostic value of the optimal model.Results The composite model showed higher prognostic performance in the training(C-index 0.791,95%CI 0.702-0.879,area under the curve,AUC=0.882)and validation groups(C-index 0.770,95%CI 0.650-0.889,AUC=0.720)than that of the clinical and radiomics models alone.The calibration curves showed good consistency between the composite model in predicting the third year probability of EFS and the actual outcome,and the survival curves showed that the EFS of the high-risk group were significantly lower than that of the low-risk group.Conclusion Composite model based on radiomics and clinical parameters can provide more comprehensive prognostic information and improve diagnostic accuracy.The nomogram provides a non-invasive diagnostic

关 键 词：淋巴瘤 结外NK-T细胞 正电子发射断层摄影术 氟脱氧葡萄糖F18 影像组学 预后 

分 类 号：R730.44[医药卫生—肿瘤] R733.1[医药卫生—临床医学]