摘要
肺结节的检测是一项非常重要的计算机辅助诊断工作。文章提出了一种多片染色重叠图像处理方法,来增强肺结节与其他健康组织间的差异性,并基于深度学习算法进行肺结节检测实验。实验使用了LIDC-IDRI数据集中的10000张肺部健康组织ROI和12000张肺结节ROI作为训练样本集,使用Alex Net卷积神经网络作为深度学习的算法网络,通过LIDC数据库中176个病人的CT图像测试,得到了95.0%的敏感性和平均5.62的假阳性率结果。实验结果表明,所提出的方法比特征提取方法等传统方法能提高肺结节的检出率。
The detection of pulmonary nodules is a very important part of computer-aided diagnosis. In this paper, a method of multi-slices coloring and superposition is proposed to increase the difference between pulmonary nodules and other healthy tissues and detect the nodules based on deep learning. The experiment uses 10000 patches of healthy tissues and 12000 patches of pulmonary nodules in LIDC-IDRI dataset as training dataset and Alex Net convolutional neural network as deep learning network.The prediction model after training is tested on 176 patients' CT images and gains the sensitive of 95.0% and an average of 5.62 false positive rates. The experimental results show that the proposed method can improve the detection rate of pulmonary nodules compared with some traditional feature extraction methods.
出处
《计算机时代》
2018年第2期5-9,共5页
Computer Era
基金
上海市临床技能与临床创新三年行动计划(16CR2042B)
国家自然科学基金(61571290
91438120
61431008)
关键词
计算机辅助诊断
深度学习
肺结节
卷积神经网络
computer-aided diagnosis
deep learning
pulmonary nodules
convolutional neural network
