工程科学学报.第43卷.第9期：1166-1173.2021年9月 Chinese Journal of Engineering,Vol.43,No.9:1166-1173,September 2021 https://doi.org/10.13374/j.issn2095-9389.2021.01.14.007;http://cje.ustb.edu.cn 基于切面识别的房间隔缺损智能辅助诊断 张文静”，李文秀，刘爱军)，武兴坤引，李剑峰)，罗涛)四 1)北京邮电大学北京先进信息网络实验室，北京1008762)首都医科大学附属北京安贞医院儿童心血管病中心，北京1000293)北京邮 电大学网络体系构建与融合北京市重点实验室，北京100876 ☒通信作者，E-mail:tuo@bupt.edu.cn 摘要针对超声心动图像质量差、噪声多，传统卷积神经网铬架构对超声心动图像的学习能力有限、表达不充分的缺点， 提出了一种基于标准切面识别的房间隔缺损(Atrial septal defect,.ASD)智能辅助诊断模型.该模型通过对超声心动图像进行 切面识别，充分融合其不同切面的语义特征，使得诊断的准确率得到明显提升。此外，还对其进行双边滤波保边去噪，并基于 此模型搭建房间隔缺损智能辅助诊断系统（简称ASD辅助诊断系统）.结果表明.该ASD辅助诊断系统的准确率高达 97.8%.且与传统卷积神经网络相比大大降低了假阴性率. 关键词深度学习：超声心动图：房间隔缺损：切面识别：双边滤波 分类号R318 Intelligent auxiliary diagnosis of atrial septal defect based on view classification ZHANG Wen-jing.LI Wen-xi,LIU Ai-ju.WU Xing-kun,LI Jian-feng,LUO Tao 1)Beijing Laboratory of Advanced Information Networks,Beijing University of Posts and Telecommunication,Beijing 100876,China 2)Pediatric Cardiovascular Center,Beijing Anzhen Hospital Affiliated to Capital Medical University,Beijing 100029,China 3)Beijing Key Laboratory of Network System Architecture and Convergence,Beijing University of Posts and Telecommunication,Beijing 100876,China Corresponding author,E-mail:tluo@bupt.edu.cn ABSTRACT Atrial septal defect (ASD)is common congenital heart disease.The detection rate of congenital heart disease has increased year by year,and ASD accounted for the largest proportion of it,reaching 37.31%.The ASD patient will suffer from shortness of breath,palpitation,weakness,etc.,with symptoms worsening with advanced age.The ASD patient will not suffer from congenital heart disease if their condition is diagnosed early.Echocardiography is a powerful and cost-effective means of detecting ASD.However, the disadvantages of echocardiography,such as noise and poor imaging quality,cause misdiagnosis of ASD.Hence,research into echocardiography-based efficient and effective detection of ASD with a deep neural network is of great significance.For echocardiography is noisy and fuzzy,and the learning and feature expression ability of the traditional convolutional neural network architecture is limited,a feature view classification based atrial septal defect intelligent auxiliary diagnostic model architecture was proposed.The different views of echocardiography possess different features,demanding more precise model extraction and combined features from echocardiography.The proposed model architecture integrates the semantic characteristics of several views,significantly improving the accuracy of diagnosis.In addition,with the aim of denoising and preserving edges,a bilateral filtering algorithm was performed.Furthermore,an ASD intelligent auxiliary diagnostic system was built based on the proposed model.The results show that the accuracy of the ASD auxiliary diagnostic system reaches 97.8%,and the false-negative rate is greatly reduced compared with the traditional convolutional neural network architecture. KEY WORDS deep learning;echocardiography:atrial septal defect;view classification;bilateral filtering 收稿日期：2021-01-14 基金项目：国家自然科学基金资助项目(61571065)基于切面识别的房间隔缺损智能辅助诊断 张文静1)，李文秀2)，刘爱军2)，武兴坤3)，李剑峰3)，罗    涛1) 苣 1) 北京邮电大学北京先进信息网络实验室，北京 100876    2) 首都医科大学附属北京安贞医院儿童心血管病中心，北京 100029    3) 北京邮 电大学网络体系构建与融合北京市重点实验室，北京 100876 苣通信作者，E-mail: tluo@bupt.edu.cn 摘    要    针对超声心动图像质量差、噪声多，传统卷积神经网络架构对超声心动图像的学习能力有限、表达不充分的缺点， 提出了一种基于标准切面识别的房间隔缺损（Atrial septal defect，ASD）智能辅助诊断模型. 该模型通过对超声心动图像进行 切面识别，充分融合其不同切面的语义特征，使得诊断的准确率得到明显提升. 此外，还对其进行双边滤波保边去噪，并基于 此模型搭建房间隔缺损智能辅助诊断系统（简称 ASD 辅助诊断系统）. 结果表明，该 ASD 辅助诊断系统的准确率高达 97.8%，且与传统卷积神经网络相比大大降低了假阴性率. 关键词    深度学习；超声心动图；房间隔缺损；切面识别；双边滤波 分类号    R318 Intelligent auxiliary diagnosis of atrial septal defect based on view classification ZHANG Wen-jing1) ，LI Wen-xiu2) ，LIU Ai-jun2) ，WU Xing-kun3) ，LI Jian-feng3) ，LUO Tao1) 苣 1) Beijing Laboratory of Advanced Information Networks, Beijing University of Posts and Telecommunication, Beijing 100876, China 2) Pediatric Cardiovascular Center, Beijing Anzhen Hospital Affiliated to Capital Medical University, Beijing 100029, China 3) Beijing Key Laboratory of Network System Architecture and Convergence, Beijing University of Posts and Telecommunication, Beijing 100876, China 苣 Corresponding author, E-mail: tluo@bupt.edu.cn ABSTRACT    Atrial  septal  defect  (ASD)  is  common  congenital  heart  disease.  The  detection  rate  of  congenital  heart  disease  has increased year by year, and ASD accounted for the largest proportion of it, reaching 37.31%. The ASD patient will suffer from shortness of breath, palpitation, weakness, etc., with symptoms worsening with advanced age. The ASD patient will not suffer from congenital heart disease if their condition is diagnosed early. Echocardiography is a powerful and cost-effective means of detecting ASD. However, the  disadvantages  of  echocardiography,  such  as  noise  and  poor  imaging  quality,  cause  misdiagnosis  of  ASD.  Hence,  research  into echocardiography-based  efficient  and  effective  detection  of  ASD  with  a  deep  neural  network  is  of  great  significance.  For echocardiography  is  noisy  and  fuzzy,  and  the  learning  and  feature  expression  ability  of  the  traditional  convolutional  neural  network architecture  is  limited,  a  feature  view  classification  based  atrial  septal  defect  intelligent  auxiliary  diagnostic  model  architecture  was proposed. The different views of echocardiography possess different features, demanding more precise model extraction and combined features from echocardiography. The proposed model architecture integrates the semantic characteristics of several views, significantly improving the accuracy of diagnosis. In addition, with the aim of denoising and preserving edges, a bilateral filtering algorithm was performed. Furthermore, an ASD intelligent auxiliary diagnostic system was built based on the proposed model. The results show that the accuracy of the ASD auxiliary diagnostic system reaches 97.8%, and the false-negative rate is greatly reduced compared with the traditional convolutional neural network architecture. KEY WORDS    deep learning；echocardiography；atrial septal defect；view classification；bilateral filtering 收稿日期: 2021−01−14 基金项目: 国家自然科学基金资助项目（61571065） 工程科学学报，第 43 卷，第 9 期：1166−1173，2021 年 9 月 Chinese Journal of Engineering, Vol. 43, No. 9: 1166−1173, September 2021 https://doi.org/10.13374/j.issn2095-9389.2021.01.14.007; http://cje.ustb.edu.cn