CT放射组学结合预测肺浸润性粘液腺癌的研究  

CT Radiomics Combined with Clinicopathological Features to Predict Invasive Mucinous Adenocarcinoma in Patients with Lung Adenocarcinoma

作　　者：张学林[1] 王杨 卓诗宇 ZHaNG Xue-lin;WANG Yang;ZHUO Shi-yu(Department of Radiation Oncology,The Second Affiliated Hospital of Hebei North University,Zhangjiakou 075100,Hebei Province,China)

机构地区：[1]河北北方学院附属第二医院肿瘤放疗科,河北张家口075100

出　　处：《中国CT和MRI杂志》2025年第2期77-79,共3页Chinese Journal of CT and MRI

基　　金：河北省卫生健康委科研基金项目(20231469);张家口市重点研发计划项目(2221174D)。

摘　　要：目的本研究旨在利用临床参数、放射学特征以及两者组合来建立和验证肺腺癌患者肺侵袭性粘液腺癌(invasive mucinous adenocarcinoma,IMA)的预测模型。方法回顾性分析了2018年8月至2023年8月的100例IMA患者和100例非IMA患者资料,用倾向性评分匹配配对。从增强CT中提取900个放射学特征,按7:3比例随机分患者为训练组和试验组,用LASSO算法选择特征,建立最优放射组学评分模型。同时,用Logistic回归建立临床模型。最后,结合两者建立组合模型,以ROC-AUC和决策曲线分析评价预测价值。结果临床和放射组学模型均表现优异,联合模型更优(P分别为0.018和0.020)。训练组和测试组中,综合模型的ROC-AUC分别为0.840和0.850,对IMA预测性能良好,Brier分数分别为0.161和0.154。结论结合放射学CT特征和临床预测因素的联合模型可能有可能预测肺癌患者的IMA。Objective The aim of this study was to establish and validate a predictive model of invasive mucinous adenocarcinoma(IMA)in patients with lung adenocarcinoma using clinical parameters,radiological features,and a combination of both.Methods Data of 100 IMA patients and 100 non-IMA patients from August 2018 to August 2023 were retrospectively analyzed and matched with propensity score.900 radiological features were extracted from enhanced CT,and patients were randomly divided into training group and experimental group according to 7:3 ratio.LASSO algorithm was used to select features and establish the optimal radiomic scoring model.Meanwhile,Logistic regression was used to establish the clinical model.Finally,a combination model is established to evaluate the prediction value by ROC-AUC and decision curve analysis.Results Both clinical and radiomic models performed well,and the combined model was superior(P 0.018 and 0.020,respectively).In the training group and the test group,the ROC AUC of the comprehensive model was 0.840 and 0.850,respectively,and the prediction performance of IMA was good,and the Brier scores were 0.161 and 0.154,respectively.Conclusion A combined model combining radiological CT features and clinical predictors may have the potential to predict IMA in patients with lung cancer.

关 键 词：肺侵袭性粘液腺癌 CT放射组学 LOGISTIC回归分析 

分 类 号：R563[医药卫生—呼吸系统]