基于CNN-L1/L2-ELM混合架构的肺结节分类研究

The CNN-L1/L2-ELM Hybrid Architecture Used to Classify Pulmonary Nodules

本文提出了一种卷积神经网络(CNN)和正则化极限学习机(L1/L2-ELM)的混合结构的分类算法,用来提高胸部CT影像中肺结节分类的准确性和分类效率.通过这两类算法的协同作用对肺结节良恶性进行有效分类.利用LIDC公开数据库进行实验验证,结果显示,本文算法所得肺结节分类正确率可达92.87%,且敏感性和特异性也分别达到87.15%和94.45%,表明本文算法对良恶性肺结节分类是有效的,且结果优于卷积神经网络和其他方法.

In this paper,a hybrid structure classification algorithm based on the convolutional neural network(CNN)and regularized extreme learning machine(L1/L2-ELM)is proposed to improve the accuracy and efficiency of pulmonary nodule classification in chest CT images.The synergistic effect of these two algorithms is used to classify benign and malignant pulmonary nodules effectively.Based on the open LIDC database,the experimental results show that,compared with the convolutional neural network and other methods,the accuracy of pulmonary nodule classification reached 92.87%and the sensitivity and specificity reached 87.15%and 94.45%,respectively.It shows that the algorithm is very effective for the classification of benign and malignant nodules.