摘要
目的探讨基于深度学习的人工智能辅助诊断肺腺癌磨玻璃结节(GGN)CT特征定量参数分析预判其病理侵袭性的价值。方法回顾2018年1月至2019年8月经手术病理证实为肺腺癌GGN54枚,按照病理结果将其分为非浸润性腺癌组(n=21)和浸润性腺癌组(n=33),并采用受试者工作特征曲线评估各项定量参数预测肺腺癌浸润性的诊断效能。将其原始数据导入人工智能软件工作站,记录其定量特征参数值。结果两组在平均CT值、3D长径、最大面面积、表面积及体积方面比较差异有统计学意义(P<0.05);两组其他定量参数差异无统计学意义(P>0.05)。平均CT值、3D长径、最大面面积、表面积及体积判断是否为浸润性腺癌组的最佳临界值分别为-476.73 Hu、10.35 mm、79.78 mm2、365.22 mm2、641.7 mm3。结论基于深度学习的人工智能CT定量参数对GGN样肺腺癌侵袭性的预判具有一定价值,平均CT值、3D长径、体积、最大面面积及表面积可作为有效的影像学预测标志物。
Objective To explore the value of deep learning-based artificial intelligence CT features quantitative parameters analysis in the prediction of the pathological invasiveness of ground-glass nodules(GGN)in lung adenocarcinoma.Methods A total of 54 ground-glass lung nodules confirmed to be lung adenocarcinoma by surgery or puncture pathology from January 2018 to August 2019 were reviewed and divided into non-invasive adenocarcinoma group(n=21)and invasive adenocarcinoma group(n=33)according to the pathological results.The diagnostic efficacy of various quantitative parameters in predicting lung adenocarcinoma invasiveness was assessed by receiver operating characteristic curve.The raw data was imported into the artificial intelligence software workstation,and the quantitative characteristic parameter values were recorded.Results There were statistically significant differences in average CT value,3D long diameter,maximum surface area,surface area and volume between the two groups(P<0.05).There was no statistically significant difference in other quantitative parameters between the two groups(P>0.05).The optimal critical values of mean CT value,3D long diameter,maximum surface area,surface area and volume for judging the invasiveness of lung adenocarcinoma were-476.73 Hu,10.35 mm,79.78 mm2,365.22 mm2,641.7 mm3,respectively.Conclusion The artificial intelligence CT quantitative parameters based on deep learning has certain value in predicting the invasiveness of ground-glass nodular lung adenocarcinoma.Mean CT value,3D long diameter,volume,maximum surface area,and surface area can be used as effective imaging predictors.
作者
周围
胡富碧
刘亚斌
朱静
吴少平
Zhou Wei;Hu Fubi;Liu Yabin;Zhu Jing;Wu Shaoping(Department of Radiology, The First Af filiated Hospital of Chengdu Medical College, Chengdu 610500, China)
出处
《成都医学院学报》
CAS
2021年第1期50-53,共4页
Journal of Chengdu Medical College
基金
成都医学院科研创新团队项目(No:CYTD19-03)。
关键词
人工智能
磨玻璃结节
肺腺癌
侵袭性
Artificial intelligence
Ground-glass nodule
Lung adenocarcinoma
Invasiveness
