[1]南嘉格列,李锐,王海霞,等.基于深度学习的肝包虫病超声图像分型研究[J].深圳大学学报理工版,2019,36(No.6(599-724)):702-708.[doi:10.3724/SP.J.1249.2019.06702]
 NANJIA Gelie,LI Rui,WANG Haixia,et al.Ultrasound image classification for hepatic echinococcosis using deep learning[J].Journal of Shenzhen University Science and Engineering,2019,36(No.6(599-724)):702-708.[doi:10.3724/SP.J.1249.2019.06702]
点击复制

基于深度学习的肝包虫病超声图像分型研究()
分享到:

《深圳大学学报理工版》[ISSN:1000-2618/CN:44-1401/N]

卷:
第36卷
期数:
2019年No.6(599-724)
页码:
702-708
栏目:
电子与信息科学
出版日期:
2019-11-20

文章信息/Info

Title:
Ultrasound image classification for hepatic echinococcosis using deep learning
文章编号:
201906016
作者:
南嘉格列1李锐2王海霞2周旭2王毅2 倪东2
1)青海大学附属医院超声科,青海西宁 810001;2)深圳大学医学超声图像计算实验室,深圳大学生物医学工程学院,医学超声关键技术国家地方联合工程实验室,广东深圳518060
Author(s):
NANJIA Gelie1 LI Rui2 WANG Haixia2 ZHOU Xu2 WANG Yi2 and NI Dong2
1) Department of Ultrasound, Qinghai University Affiliated Hospital, Xining 810000, Qinghai Province, P.R.China 2) Lab of Medical UltraSound Image Computing, MUSIC, National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, School of Biomedical Engineering, Health Science Center, Shenzhen University, Shenzhen 518060, Guangdong Province, P.R.China
关键词:
生物医学工程肝包虫病卷积神经网络超声图像病灶分型视觉注意力模型度量学习迁移学习双分支分类网络
Keywords:
artificial intelligence hepatic echinococcosis convolutional neural network ultrasonoscopy lesions classification visual attention model metric learning transfer learning dual-branch classification network
分类号:
R318; TP751
DOI:
10.3724/SP.J.1249.2019.06702
文献标志码:
A
摘要:
肝包虫病是一种严重的地域性寄生虫病,其病灶分型主要依靠临床医生对超声图像的主观判断,疾病筛查工作中十分耗时,且容易造成误判.提出一种基于超声图像的肝包虫病病灶智能分型方法,首先从肝脏包虫病超声图像中直接裁剪得到病灶区域图像,利用深度卷积神经网络(convolutional neural network, CNN)提取图像多尺度特征,然后结合视觉注意力模型,通过分类网络的主分支和辅助分支分别学习图像的整体和局部细节特征,最后使用度量学习来表征同类别之间样本的相似特征,实现对9种类型的包虫病病灶进行全自动分类.构建了一个18层CNN网络,通过7 000张图像完成训练,在2 000张图像上测试得到的平均准确率为82%,平均F1分数为82%.实验结果充分表明,该方法能够有效应用于肝包虫病超声图像分型.
Abstract:
Hepatic echinococcosis is a serious regional parasitic disease, and its lesions classification mainly rely on the subjective judgment of clinicians on ultrasonic images, which is time-consuming and easy to cause misjudgment in disease screening. This study proposed an intelligent classification method for hepatic echinococcosis lesions based on ultrasound images. First of all, lesion images are cropped from the hepatic echinococcosis ultrasound images directly. The deep convolutional neural network (CNN) is used to extract multi-scale features. And then based on visual attention model, the main branch and the auxiliary branch of the classification network learn global image features and details local features, respectively. Finally, metric learning is used to measure the similarity of the samples in the same category. Automatic classification of nine types of echinococcosis lesions are realized. An 18-layer CNN network was constructed, and the training was completed with 7000 images. The average accuracy rate of the test on 2000 images was 82%, and the average F1 score was 82%. The results demonstrate that the proposed method can be effectively applied to the ultrasound image classification of hepatic echinococcosis.

相似文献/References:

[1]潘晓畅,冯乃章,陈思平.二维Savitzky-Golay数字差分器剪切波弹性成像方法[J].深圳大学学报理工版,2018,35(No.1(001-110)):22.[doi:10.3724/SP.J.1249.2018.01022]
 PAN Xiaochang,FENG Naizhang,et al.A shear wave elastography method based on 2D Savitzky-Golay digital differentiator[J].Journal of Shenzhen University Science and Engineering,2018,35(No.6(599-724)):22.[doi:10.3724/SP.J.1249.2018.01022]
[2]刘刚,钱建庭,李先明,等.X射线光声成像的信号检测与仿真[J].深圳大学学报理工版,2018,35(No.3(221-330)):324.[doi:10.3724/SP.J.1249.2018.03324]
 LIU Gang,CHIN C T,LI Xianming,et al.Simulation and experimental detection of X-ray photo-acoustic[J].Journal of Shenzhen University Science and Engineering,2018,35(No.6(599-724)):324.[doi:10.3724/SP.J.1249.2018.03324]
[3]胡鹏辉,王娜,王毅,等.基于全卷积神经网络的肛提肌裂孔智能识别[J].深圳大学学报理工版,2018,35(No.3(221-330)):316.[doi:10.3724/SP.J.1249.2018.03316]
 HU Penghui,WANG Na,WANG Yi,et al.Automatic recognition of levator hiatus based on fully convolutional neural networks[J].Journal of Shenzhen University Science and Engineering,2018,35(No.6(599-724)):316.[doi:10.3724/SP.J.1249.2018.03316]

备注/Memo

备注/Memo:
Received:2019-03-20;Accepted:2019-05-13 Foundation:National Natural Science Foundation of China ( 61571304 ) Corresponding author:Professor NI Dong.E-mail: nidong@szu.edu.cn Citation:NANJIA Gelie, LI Rui, WANG Haixia, et al. Ultrasound image classification for hepatic echinococcosis using deep learning[J]. Journal of Shenzhen University Science and Engineering, 2019, 36(6): 702-708.(in Chinese)
更新日期/Last Update: 2019-11-30