摘要
目前肺炎类型判别主要依靠医生的经验,但一些肺炎的CT影像极为近似,即使有经验的医生,也容易判别错误,造成误诊;为此提出卷积神经网络分类算法,该算法由3个卷积层、3个亚采样层及1个完全连接层组成,并且对卷积层进行了特殊结构处理,由反向传播算法调整网络参数,并对反向传播过程提出了改进;临床实验证明,该方案较现在普遍研究的分类算法,如adaboost算法和svm算法具有更高的识别率和准确度,并且改进的卷积神经网络防止了训练数据时过拟合现象的产生。
Currently, the diagnosis of pneumonia type mainly depends on the experience of doctors. But some of the CT images of pneu- monia were very similar. It is easy to misdiagnose, even experienced doctors. Therefore, a classification algorithm based on convolution neu- ral network is proposed, which consisted of three convolution layers, three subsampling layers and one fully connected layer , and also a spe- cial structural processing is proceeded on the convolution layer. The network parameters are regulated through back propagation algorithm, which can also improve the back propagation process. The clinical experimental results show that the algorithm can accurately classify CT images of different pneumonia than general research recognition algorithm, such as Adaboost and SVM algorithms, and the revised convolution neural network can prevent over--fitting phenomena in training data.
出处
《计算机测量与控制》
2017年第4期185-188,共4页
Computer Measurement &Control
基金
国家自然基金项目(61403109)
黑龙江省自然科学基金(F201240)
黑龙江省教育厅科技研究项目(12531571)
关键词
肺炎判别
CT影像
深度学习
细微特征差异
recognition of pneumonia type
CT image
deep learning
subtle differences
