CT特征联合人工智能定量参数评估ⅠA期肺腺癌高级别组织学亚型  

CT characteristics combined with artificial intelligence quantitative parameters for evaluating high-grade histologic subtype in stageⅠA lung adenocarcinoma

作　　者：梁演婷 林欢[2,3,4] 李夙芸 刘晨 莫梓阳 刘昱琳 刘再毅 LIANG Yanting;LIN Huan;LI Suyun;LIU Chen;MO Ziyang;LIU Yulin;LIU Zaiyi(Guangdong Cardiovascular Institute,Guangzhou 510080,China;Department of Radiology,Guangdong Provincial People's Hospital,Guangdong Academy of Medical Sciences,Guangzhou 510080,China;Guangdong Provincial Key Laboratory of Artificial Intelligence in Medical Image Analysis and Application,Guangzhou 510080,China;School of Medicine,South China University of Technology,Guangzhou 510006,China)

机构地区：[1]广东省心血管病研究所,广东广州510080 [2]广东省人民医院(广东省医学科学院)放射科,广东广州510080 [3]广东省医学影像智能分析与应用重点实验室,广东广州510080 [4]华南理工大学医学院,广东广州510006

出　　处：《中国医学影像技术》2023年第2期199-203,共5页Chinese Journal of Medical Imaging Technology

基　　金：国家重点研发计划项目(2021YFF1201003)。

摘　　要：目的探讨CT特征联合人工智能(AI)定量参数评估临床ⅠA期高级别肺腺癌(LADC)的价值。方法纳入482例ⅠA期LADC患者,以病理分级系统将其分为低级别组(n=366)和高级别组(n=116);比较组间临床及影像学主观特征及AI定量参数差异,以logistic回归分析筛选评估高级别LADC的独立因子,并构建主观特征模型、AI模型及联合模型;绘制受试者工作特征曲线,评估各模型诊断效能。结果对于评估高级别LADC,主观特征模型中的结节类型、实性成分占比及空气支气管征,AI模型的CT平均值及峰度,以及联合模型中的AI评分、实性成分占比及空气支气管征均为独立预测因子(P均<0.05)。主观特征模型、AI模型及联合模型评估高级别LADC的曲线下面积分别为0.886、0.885及0.901,联合模型评估效能优于主观特征模型及AI模型(P均<0.05)。结论CT特征联合AI定量参数有助于术前无创评估临床ⅠA期高级别LADC。Objective To explore the value of CT characteristics combined with artificial intelligence(AI)quantitative parameters for evaluating high-grade histologic subtype in clinical stageⅠA lung adenocarcinoma(LADC).Methods Totally 482 patients with stageⅠA LADC were enrolled and divided into low-grade group(n=366)and high-grade group(n=116)according to pathological grading system.Clinical and imaging subjective characteristics and AI quantitative parameters were compared between groups.The independent predictors of high-grade LADC were selected using logistic regression analysis,and the subjective characteristic model,the AI model and the combined model were constructed,respectively.Receiver operating characteristic curves were plotted to evaluate the performance of each model.Results Nodule type,consolidation-to-tumor ratio and air bronchogram sign in subjective characteristic model,the mean CT value and kurtosis in AI model,as well as AI-score,consolidation-to-tumor ratio and air bronchogram sign in the combined model were all independent predictors for evaluating high-grade LADC(all P<0.05).The area under the curve for predicting high-grade LADC of the subjective characteristics model,AI model and combined model was 0.886,0.885 and 0.901,respectively.The combined model performed better than the subjective characteristics model and AI model(both P<0.05).Conclusion CT characteristics combined with AI quantitative parameters were helpful for preoperatively and non-invasively evaluation on high-grade histologic subtype in clinical stageⅠA LADC.

关 键 词：肺肿瘤 体层摄影术 X线计算机 病理学 人工智能 

分 类 号：R734.2[医药卫生—肿瘤] R814.42[医药卫生—临床医学]