人工智能体积密度法判断肺亚实性结节的浸润性研究

Evaluation of the Invasion of Pulmonary Subsolid Nodules by the Artificial Intelligence Volumetric Density Method

目的:探讨人工智能(AI)体积密度法判断肺亚实性结节(SSNs)浸润性的价值。方法:回顾性分析106例患者的108枚SSNs的CT和病理结果,将结节分为腺体前驱病变组和腺癌组。通过肺结节AI软件测量并比较两组的最大CT值、最小CT值、平均CT值、峰度、偏度、Perc.25%、Perc.50%、Perc.75%、Perc.95%、结节体积、结节平均径等CT定量参数。使用Medcalc软件得出受试者工作特征曲线(ROC),评价诊断SSNs浸润性的敏感度、特异度、阳性预测值及阴性预测值,用逻辑回归分析评估他们的诊断性能。结果:SSNs的多数CT定量参数差异存在统计学意义,其中,诊断效能最高的是Perc.25%,AUC达0.797;其次为Perc.50%和平均CT值,AUC均为0.787。Logistic回归分析显示,将诊断效能最高的Perc.25%分别与Perc.50%和平均CT值两两建立联合诊断模型1,其中Perc.25%与平均CT值的模型诊断效能最高,且联合诊断模型诊断效能高于Perc.25%与平均CT值单独的诊断效能。Medcalc软件分析显示,Perc.25%≥-578 HU和平均CT值≥-468 HU的SSNs病理表现为腺癌的可能性大。将Perc.25%与结节平均径结合,可获得对判断SSNs浸润性非常有价值的联合诊断模型2。结论:AI体积密度法对SSNs的浸润性有较高的诊断价值,联合使用Perc.25%与平均CT值比单独使用更能准确地判断浸润性。

Objective: To explore the value of the artificial intelligence(AI) volumetric density method in determining the invasion of pulmonary hyposolid nodules(SSNs). Methods: A total of 108 SSNs and the pathological results of 106 patients were reviewed, and these were divided into a glandular prodromal lesions group and an adenocarcinoma group. Pulmonary nodule AI software was used to measure and compare the CT quantitative parameters of the two groups, including the maximum CT value, minimum CT value, average CT value, kurtosis, skewness, Perc.25%, Perc.50%, Perc.75%, Perc.90%,nodule volume, and mean nodule diameter. Moreover, a receiver operating characteristic curve(ROC) was obtained by MedCalc software to evaluate the sensitivity, specificity, positive predictive value, and negative predictive value for the diagnosis of SSN infiltration, and their diagnostic performance was evaluated by logistic regression analysis. Results: There were significant differences in most CT quantitative parameters of SSNs. The highest diagnostic efficiency was Perc.25% and the AUC was 0.797, while the AUC was 0.787 for Perc.50% and the mean CT value. Logistic regression analysis showed that Perc.25% with the highest diagnostic efficiency was combined with Perc.50% and the mean CT value. The model with Perc.25% and the mean CT value had the highest diagnostic efficiency, and the combined diagnostic model had a higher diagnostic efficiency than Perc.25% and the mean CT value alone. According to MedCalc software, SSNs with Perc.25% ≥-578 HU and mean CT values ≥-468 HU were more likely to be in the adenocarcinoma group. In this study, Perc.25% was combined with the mean diameter of nodules, and a very valuable combined diagnostic model II was obtained to judge the infiltration of SSNs. Conclusion: The AI volume density method has a high diagnostic value for SSN invasion. Moreover, the combination of Perc.25% and mean CT value can accurately judge the invasion than the use of average CT value alone,providing a quantitative basis for the clinical management of SSNs.