人工智能定量参数在直径≤2 cm磨玻璃密度肺腺癌浸润程度中的预测价值  

Predictive value of artificial intelligence quantitative parameters in the degree of invasion of lung adenocarcinoma with a glass density of less than 2 cm in diameter

作　　者：李善杰 陈阳阳[2] 刘国华 Li Shanjie;Chen Yangyang;Liu Guohua(Imaging Department,Xinxiang Alliance Hospital,Henan 453800,China;不详)

机构地区：[1]新乡同盟医院影像科,453800 [2]新乡医学院第一附属医院影像科 [3]新乡同盟医院胸外科,453800

出　　处：《实用医学影像杂志》2025年第1期25-28,共4页Journal of Practical Medical Imaging

摘　　要：目的分析人工智能定量参数在直径≤2 cm磨玻璃密度肺腺浸润程度中的预测价值。方法随机选取93例自2021年8月至2023年12月在本院进行诊断与治疗的肺癌患者,微浸润腺癌及原位癌患者共59例纳入对照组,浸润性腺癌34例纳入研究组。应用CT扫描仪进行胸部CT扫描,将胸部CT肺窗薄层图像导入人工智能肺结节辅助诊断系统中,分析2组患者磨玻璃密度结节检出情况以及肺结节手术时间间隔,比较2组人工智能定量参数,分析直径≤2 cm磨玻璃密度肺腺癌侵袭性影响因素,2组人工智能定量参数ROC曲线及各项参数预测直径≤2 cm磨玻璃结节的灵敏度及特异度。结果全部107个磨玻璃密度结节中共计检出混合磨玻璃结节(mGGN)68个及纯磨玻璃结节(pGGN)39个,术后病理诊断结果表明观察组共计检出结节45个,对照组共计检出结节62个,与对照组相比,观察组患者pGGN占比更低,mGGN占比更高(P<0.05),2组患者肺结节手术时间间隔差异无统计学意义(P>0.05)。对照组与观察组患者最小CT值差异无统计学意义(P>0.05),与对照组相比,观察组患者年龄更高,最大CT值、平均CT值、最大面面积、三维长径及体积均更高(P<0.05)。多因素Logistic回归分析结果表明最大CT值及三维长径为直径≤2 cm磨玻璃密度肺腺癌侵袭性影响因素(P<0.05)。各参数在直径≤2 cm磨玻璃密度肺腺癌中的诊断效能对比可知,联合诊断效能最高,其次为三维长径、体积、最大面面积、最大CT值以及平均CT值,联合诊断灵敏度高于最大CT值、平均CT值(P<0.05)。结论人工智能定量参数可准确预测直径≤2 cm磨玻璃密度肺腺癌浸润程度,以最大CT值与三维长径建立的联合模型具有最高的诊断效能。Objective To analyze the predictive value of artificial intelligence quantitative parameters in the degree of lung gland infiltration with a diameter≤2 cm ground glass density.Methods Ninety-three patients with lung cancer diagnosed and treated in our hospital from August 2021 to December 2023 were randomly selected to perform chest CT scanning with a CT scanner,and the thin layer images of chest CT lung Windows were imported into the artificial intelligence lung nodule auxiliary diagnosis system.The detection of ground glass density nodules and the interval of lung nodule operation were analyzed in the two groups.The artificial intelligence quantitative parameters in the two groups were compared to analyze the factors affecting the invasiveness of lung adenocarcinoma with ground glass density≤2 cm in diameter,the Receiver operator characteristic curve(ROC)of the two groups′artificial intelligence quantitative parameters and the sensitivity and specificity of each parameter in predicting ground glass nodule with a diameter≤2 cm.Results A total of 68 mixed ground-glass nodules(mGGN)and 39 pure ground-glass nodules(pGGN)were detected in all 107 ground-glass density nodules.Postoperative pathological diagnosis results showed that 45 nodules were detected in the observation group and 62 nodules were detected in the control group.Compared with the control group,the proportion of PGGN in the observation group was significantly lower and the proportion of mGGN was significantly higher(P<0.05).There was no significant difference in the time interval of pulmonary nodule operation between the two groups(P>0.05).There was no significant difference in the minimum CT value between the control group and the observation group(P>0.05).Compared with the control group,the patients in the observation group were older,and the maximum CT value,average CT value,maximum surface area,three-dimensional length diameter,and volume were higher(P<0.05).The results of multivariate Logistic regression analysis showed that the maximum C

关 键 词：肺腺癌 人工智能 磨玻璃密度 预测价值 诊断效能 

分 类 号：R73[医药卫生—肿瘤]