人工智能联合循环染色体异常细胞在鉴别亚实性结节型肺腺癌的病理亚型的价值  

Diagnostic value of artificial intelligence combined with circulating chromosomal abnormal cells in pathological subtypes of subsolid nodular lung adenocarcinoma

作　　者：乔姣姣 张国瑞[1] 陈红杰[1] 余亚丽 刘淼 刘红[1] QIAO Jiaojiao;ZHANG Guorui;CHEN Hongjie;YU Yali;LIU Miao;LIU Hong(Department of Respiratory and Critical Care Medicine,the First Affiliated Hospital of Zhengzhou University,Zhengzhou,Henan 450000,China)

机构地区：[1]郑州大学第一附属医院呼吸与危重症医学科,河南郑州450000

出　　处：《临床肺科杂志》2025年第5期661-666,共6页Journal of Clinical Pulmonary Medicine

基　　金：国家重点研发计划精准医学研究专项(2016YFC0901100);河南省重点研发专项项目(241111313100);河南省高等学校重点科研项目(24A320043)。

摘　　要：目的探讨人工智能(AI)参数联合循环染色体异常细胞(CAC)在鉴别亚实性结节型肺腺癌的病理亚型的价值。方法回顾性分析2023年3月至2024年3月在郑州大学第一附属医院收治的224例亚实性肺结节患者。根据术后病理将患者分为微浸润组(n=164)和浸润组(n=60)。将各组的CT图像导入人工智能辅助诊断系统,记录亚实性肺结节的定量参数,人工识别CT征象,同时采用荧光原位杂交技术得到CAC计数。将两组进行单因素分析,有统计学差异(P<0.05)的变量纳入多因素Logistic回归中构建预测模型,同时进行ROC曲线分析评价其性能。结果两组的年龄、结节类型、毛刺征、胸膜凹陷征、AI恶性概率、2D长径、2D短径、3D长径、3D短径、体积、平均CT值、最大CT值、标准差及CAC计数差异有统计学意义(P<0.05)。多因素Logistic回归分析显示3D长径(OR=22.618、95%CI=8.080~63.312)、最大CT值(OR=7.469、95%CI=3.124~17.853)、CAC(OR=5.545、95%CI=2.253~13.648)是预测亚实性结节型肺腺癌的病理亚型的独立危险因素。3D长径、最大CT值、CAC三者单独及联合检测的AUC分别为0.854(95%CI:0.792~0.915)、0.793(95%CI:0.725~0.861)、0.675(95%CI:0.591~0.759)、0.912(95%CI:0.866~0.958),联合检测优于单独(Z_(联合-3D长径)=3.229,P=0.001,Z_(联合-最大CT值)=4.411,P<0.001,Z_(联合-CAC=4.995),P<0.001)。结论3D长径、最大CT值、CAC是预测亚实性结节型肺腺癌病理亚型的独立危险因素,上述指标联合检测可以提高亚实性结节型肺腺癌病理亚型的鉴别能力。Objective To investigate the value of artificial intelligence(AI)parameters combined with circulating chromosomal abnormal cells(CAC)in distinguishing pathological subtypes of subsolid nodule(SSN)lung adenocarcinoma.Methods 224 patients with subsolid pulmonary nodules admitted to the First Affiliated Hospital of Zhengzhou University from March 2023 to March 2024 were retrospectively analyzed.Patients were divided into micro-infiltration group(n=164)and infiltration group(n=60)based on postoperative pathology.The CT images of each group were imported into an artificial intelligence-assisted diagnosis system to record quantitative parameters of subsolid pulmonary nodules,manually identify CT signs,and use fluorescence in situ hybridization to obtain CAC counts.Univariate analysis was performed between the two groups,and variables with statistical differences(P<0.05)were included in multivariate logistic regression to build a predictive model,and ROC curve analysis was performed to evaluate their performance.Results There were statistically significant differences between the two groups in age,nodule type,burr sign,pleural indentation sign,AI malignancy probability,2D long diameter,2D short diameter,3D long diameter,3D short diameter,volume,average CT value,maximum CT value,standard deviation and CAC count(P<0.05).Multivariate Logistic regression analysis showed that 3D long diameter(OR=22.618、95%CI=8.080-63.312),maximum CT value(OR=7.469、95%CI=3.124-17.853),and CAC(OR=5.545、95%CI=2.253-13.648)were independent risk factors for predicting the pathological subtype of subsolid nodular lung adenocarcinoma.The AUCs of 3D longitudinal diameter,maximal CT value,and CAC were 0.854(95%CI:0.792-0.915),0.793(95%CI:0.725-0.861),0.675(95%CI:0.591-0.759),and 0.912(95%CI:0.866-0.958)for the three tests individually and in combination,respectively.Combined detection was superior to separate(Z_(combined-3D-LD)=3.229,P=0.001,Z_(combined-max CT)=4.411,P<0.001,Z_(combined-CAC)=4.995,P<0.001).Conclusion 3 D long diameter,maximum C

关 键 词：肺腺癌 肺结节 病理亚型 人工智能 循环染色体异常细胞 鉴别 

分 类 号：R734.2[医药卫生—肿瘤]