放射组学在肺癌诊断中的应用

目的:通过放射组学对肺癌病例进行定量特征提取，优化选择，然后通过机器学习方法实现肺癌病例讨论和分析。方法:通过公开数据库LIDC中提取224例和医院收集250例肺结节病例，提取共841个放射组学特征；对特征进行正态分析和方差齐性分析，双独立样本t检验进行降维；其余采用秩和分析降维，之后采取Pearson相关系数降维，最后通过机器学习方法进行分类。结果:来自LIDC数据库和来自医院的数据在基于随机森林的分类器中的结果分别为AUC=0.657 1、ACC=76.26%，AUC=0.866 7、ACC=76%；在基于支持向量机的分类器中的结果分别为AUC=0.642 9，ACC=76.37%，AUC=0.773 3、ACC=72%。结论:在肺癌良恶诊断鉴别中，使用放射组学特征方法可以鉴别良恶性。基于纹理特征的计算机辅助诊断系统可以提高对此类结节的诊断效能。

The application of radiomics in the diagnosis of lung cancer

Objective: To quantitatively extract and optimizeradiomicsfeatures for lung cancer cases and to analyze and discuss lung cancer cases by machine learning method. Methods:We obtained images of 224 patients from LIDC databaseand 250 patients from hospital, and 841 radiomicsfeatures were extracted. The features were used to perform dimensionality reduction by the double independent sample t-test, when normality of distribution and Homogeneity variance were calculated. Futhermore, the dimensionality reduction was performed by the rank sum test. And then, the Pearson correlation coefficient was used for further dimensionality reduction.Finally, machine learning method was used for classification. Results: In the classifier based on the random forest, the LIDC database showed that ACC=76.26%, AUC=0.657 1 and the data from the Hospital showed that ACC=76%, AUC=0.866 7. In the classifier based on the support vector machine, the LIDC database showed that ACC=76.37%, AUC=0.642 9, and the data from the hospital showed that ACC=72%, AUC=0.773 3. Conclusion: In pulmonary nodules, radiomics can be used to identify benign and malignant nodules. Texture-based computer-aided diagnosis systems may improve the diagnostic efficacy on pulmonary nodules.