摘要
目的探讨基于人工智能学习CT密度定量对肺内孤立性亚实性及实性结节的良恶性诊断价值。方法回顾性分析延安大学附属医院2017年6月至2021年6月经手术后病理学检查证实的73例肺内孤立性亚实性及实性肺结节患者,利用人工智能肺结节辅助诊断分析软件对肺结节患者胸部CT进行定量分析。采用独立样本t检验或非参数检验比较两组患者CT密度定量参数的差异,进行单因素与多因素的二元logistic回归分析,绘制ROC曲线并计算曲线下面积(AUC),鉴别肺腺癌的独立危险因素。P<0.05为差异有统计学意义。结果肺亚实性及实性结节的大小(长、短径)及CT最大值差异均无统计学意义。CT平均值、CT最小值与CT值方差有统计学意义(P<0.05)。回归分析显示CT值方差是肺腺癌组患者独立影响因素(OR值=1.020,95%CI:1.003~1.037),预测准确率80.4%。ROC曲线下面积AUC=0.812(95%CI:0.680~0.943),临界值60.13,灵敏度78.6%,特异度78.6%。结论人工智能肺结节CT密度定量参数能够鉴别肺腺癌与肺内良性结节,CT值方差是敏感、客观的影像学定量指标。
Objective To explore the value of CT density quantification based on artificial intelligence learning in the diagnosis of subsolid and solid pulmonary nodules which is benign or malignan.Methods 73 patients with pulmonary nodules who confirmed by surgical pathology in Affiliated Hospital of Yan’an University from June 2017 to June 2021 were retrospectively collected.Quantitative analyses of chest CT in patients with pulmonary nodules were carried out using artificial intelligence pulmonary nodule auxiliary diagnosis and analysis software.Independent sample t-test or non-parametric test was used to compare the differences in quantitative parameters of CT density between two groups.Univariate and multivariate binary logistic regression analysis was performed,ROC curve was drawn and the area under the curve(AUC) was calculated to identify independent risk factors for lung adenocarcinoma.P<0.05 was considered statistically significant.Results There was no significant difference between the size(long diameter,short diameter) with CT maximum value of subsolid and solid nodules.There were significant differences between the mean value of CT,the minimum value of CT and the variance of CT(P<0.05).Regression analysis showed that the variance of CT value was an independent influenced factor of patients in lung adenocarcinoma group(OR=1.020,95%CI:1.003~1.037),and the prediction aceuracywas80.4%.The area under the ROC curve AUC=0.812(95%CI:0.680~0.943),the critical value is 60.13,the sensitivity is 78.6%,and the specificity is 78.6%.Conclusion The quantitative parameters of CT density of pulmonary nodules with artificial intelligence can distinguish pulmonary adenocarcinoma and benign pulmonary nodules.The variance of CT value is a sensitive and objective imaging quantitative index.
作者
牛媛
黄晓旗
陈新花
王剑
闫军
李建龙
NIU Yuan;HUANG Xiaoqi;CHEN Xinhua;WANG Jian;YAN Jun;LI Jianlong(Affiliated Hospital of Yan'an University,Yan'an 716000,China)
出处
《延安大学学报（医学科学版）》
2022年第1期72-76,共5页
Journal of Yan'an University:Medical Science Edition
关键词
CT密度定量
肺结节
亚实性与实性
CT density quantification
Pulmonary nodules
Subsolid and solid
