【摘要】 目的 探讨CT影像组学联合临床及CT特征构建的诊断模型对肺腺癌P53突变状态的预测价值。方法 回顾性收集177例肺腺癌患者资料,其中P53突变型75例,野生型102例。按7︰3的比例随机分为训练集(123例)和验证集(54例)。应用Ma Zad软件进行感兴趣区(ROI)勾画,采用最小绝对收缩和选择算子(LASSO)筛选特征。基于logistic回归分析构建预测模型,并评价其诊断效能。结果 训练集2组间比较发现,病灶直径、实性成分最大径、分叶征及淋巴结转移有统计学差异(P<0.05)。影像组学评分模型、临床影像特征模型及联合模型训练集曲线下面积(AUC)分别为0.778、0.759、0.817,验证集AUC分别为0.668、0.722、0.783。临床决策曲线显示联合模型具有较好的临床净收益。结论 CT影像组学、病灶直径及淋巴结转移构建的联合模型对肺腺癌P53突变状态具有较好的预测价值。更多还原

【Abstract】 Objective To investigate the predictive value of the diagnostic model constructed by CT radiomics combined with clinical and CT features for the P53 mutation status in lung adenocarcinoma.Methods The data of 177 patients with lung adenocarcinoma were retrospectively collected, including 75 cases of P53 mutant type and 102 cases of wild type.They were randomly divided into training cohort(n=123) and validation cohort(n=54) in a 7:3 ratio.The region of interest(ROI)was delineated using MaZda software, and the features were screened using the least absolute shrinkage and selection operator(LASSO).The predictive model was developed based on logistic regression analysis, and its diagnostic efficacy was evaluated. Results There were statistically significant differences in lesion diameter, maximum solid component diameter, lobulation sign, and lymph node metastasis between the two groups in the training cohort(P<0.05).The area under the curve(AUC) in the training cohort of the radiomics score model, clinical imaging feature model, and combined model were 0.778,0.759,and 0.817,and the AUC in the validation cohort were 0.668,0.722, and 0.783,respectively.The clinical decision curve suggested that the combined model demonstrated superior clinical net benefits. Conclusion The combined model constructed by CT radiomics, lesion diameter, and lymph node metastasis has better predictive value for the P53 mutation status in lung adenocarcinoma.更多还原