中文摘要：

在肺结节诊断方法研究中,传统机器学习诊断方法存在诊断性能不稳定的问题。为了提高孤立性肺结节的识别准确率,提出基于粒子群优化（particle swarm optimization,PSO）参数的极限学习机（extreme learning machine,ELM）辅助诊断方法。首先采用PSO选取ELM最佳的初始权重ω和偏置b;然后利用最佳参数ω和b对ELM进行训练;再利用PSO-ELM对通过稀疏自编码得到的肺结节特征进行分类识别。实验中,将传统机器学习算法与本文方法进行对比,结果表明,利用粒子群优化算法进行优化的极限学习机在孤立性肺结节诊断方面具有较高识别准确率和稳定的分类性能,可以作为一种有效的肺结节诊断工具。

英文摘要：

In order to improve the accuracy of the solitary pulmonary nodule recognition, a lung cancer diagnosis method based on PSO-ELM was proposed. Firstly, Particle Swarm Optimization （PSO） was used to select the best input weight to and bias b of Extreme Learning Machine （ELM）. Secondly, ELM was trained with to and b which were the best parameters selected by PSO. Finally, the PSO-ELM was used to diagnose the solitary pulmonary by features which were extracted by spare encoding. Compared with the traditional machine learning method, the recognition accuracy of Extreme Learning Machine based on particle swarm optimization is higher and the diagnosis performance is more stable, which can be used as an effective pulmonary nodules diagnosis tool.