网站首页            期刊简介             编委会             投稿指南             期刊订阅             下载中心             在线留言            联系我们             English
  2025年4月27日 星期日  
文章快速检索
中国生物医学工程学报  2018, Vol. 37 Issue (1): 17-24    DOI: 10.3969/j.issn.0258-8021. 2018. 01.003
  论著 本期目录 | 过刊浏览 | 高级检索 |
基于卷积神经网络的外周血白细胞分类
陈畅1 ,程少杰2 ,李卫滨2, 陈 敏2*
1中国科学技术大学信息科学技术学院,合肥 230022 ;
2解放军福州总医院全军检验医学研究所,福州 350025
A Peripheral Blood WBC Classification with ConvolutionalNeural Network
Chen Chang1, Cheng Shaojie2, Li Weibin2 ,Chen Min2*
1Institute of Information Science and Technology,University of Science and Technology of China,Hefei 230022, China;
2The PLA Institute of Laboratory Medicine,Fuzhou General Hospital,Fuzhou 350025, China
全文: PDF (872 KB)   HTML (1 KB) 
输出: BibTeX | EndNote (RIS)      
摘要 白细胞图像的自动分类有助于提高临床诊疗效率,但仍需进一步改进方法以提高分类正确率。探索用卷积神经网络(CNN)进行外周血白细胞图像的自动分类识别。在深度学习框架Caffe上,以AlexNet和LeNet为网络原型构建CNN训练平台;用CellaVision DM96采集外周血涂片中的5类白细胞图像,经人工鉴定后按训练∶校验∶测试=7∶2∶1的比例,随机分配图像构建原始数据集,再通过平移、旋转及镜像构建扩充数据集;训练时采用随机梯度下降算法优化模型权值,以分类准确率>95%为目标评估训练结果及优化调整网络结构。结果发现,AlexNet的训练误差无法收敛,陷入局部极小,LeNet则达到预期目标。随后对LeNet网络进行删减优化,获得一轻量高效的新结构——CCNet,其在模型大小、训练用时和分类用时上分别仅为LeNet的1/1000、1/3和1/30。两者对979张5类细胞图像的最佳分类准确率分别达到99.69%和99.18%,高于目前同类研究报道。结果表明,CNN可用于5类白细胞图像的“端对端”分类识别,特别是CCNet模型兼具准确与效率优势。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
陈畅
程少杰
李卫滨
陈 敏
关键词 深度学习卷积神经网络白细胞形态分类识别    
Abstract:The automatic classification of white blood cell (WBC) image is essential because it helps to enhance the efficiency of clinical diagnosis and treatment. However, the classification accuracy is still need to be boost for adapting to practical applications. In this paper, we proposed an automatic classification method based on the convolution neural network (CNN). We tentatively fed our training dataset into AlexNet and LeNet using a widely used deep learning platform Caffe. Five classes of WBCs images collected by a CellaVision DM96 in peripheral blood smears were adopted as the training dataset. These manual labeled images were apportioned into three groups (training, validation and testing) randomly to construct the original dataset according to the proportion of 7:2∶1. With the augmentation methods, such as rotation and mirror, we expanded the original dataset. Stochastic gradient descent algorithm was adopted as the optimizing method for training CNNs. The experimental results demonstrated that the network structure of AlexNet was unsuitable to achieve the ideal classification accuracy which more than 95%. While the network structure of LeNet had achieved the expected target. However, the more massive and more time consuming of LeNet suggested us to further optimize the connection of layers to derive a new network with lightweight structure, named as CCNet. The model size, time for training, and time for evaluation of CCNet were only 1/1000, 1/3, and 1/30 compared with LeNet, respectively. The best classification accuracy of CCNet and LeNet for five classification of WBCs was 99.69% and 99.18% with 979 WBC images, higher than those of the previous reports. It demonstrated that CNNs especially CCNet had clear advantages than previous works both in classification accuracy and speed.
Key wordsdeep learning    convolution neural network    leukocyte morphology    classification
收稿日期: 2016-10-11     
PACS:  R318  
基金资助:福建省科技重点项目(2012Y0058)
通讯作者: E-mail: fzcmin@qq.com   
引用本文:   
陈畅,程少杰,李卫滨, 陈 敏. 基于卷积神经网络的外周血白细胞分类[J]. 中国生物医学工程学报, 2018, 37(1): 17-24.
Chen Chang, Cheng Shaojie, Li Weibin,Chen Min. A Peripheral Blood WBC Classification with ConvolutionalNeural Network. Chinese Journal of Biomedical Engineering, 2018, 37(1): 17-24.
链接本文:  
http://cjbme.csbme.org/CN/10.3969/j.issn.0258-8021. 2018. 01.003     或     http://cjbme.csbme.org/CN/Y2018/V37/I1/17
版权所有 © 2015 《中国生物医学工程学报》编辑部
本系统由北京玛格泰克科技发展有限公司设计开发