基于GAN和MS-ResNet的房颤自动检测模型  

An Automatic Atrial Fibrillation Detection Model Based on GAN and MS-ResNet

作　　者：秦静[1] 韩悦 王立永 季长清[2,3] 刘璐[4] 汪祖民[2] QIN Jing;HAN Yue;WANG Liyong;JI Changqing;LIU Lu;WANG Zumin(College of Software Engineering,Dalian University,Dalian 116622,Liaoning,China;College of Information Engineering,Dalian University,Dalian 116622,Liaoning,China;College of Physical Science and Technology,Dalian University,Dalian 116622,Liaoning,China;Zhongshan Hospital Affiliated to Dalian University,Dalian 116001,Liaoning,China)

机构地区：[1]大连大学软件工程学院,辽宁大连116622 [2]大连大学信息工程学院,辽宁大连116622 [3]大连大学物理科学与技术学院,辽宁大连116622 [4]大连大学附属中山医院,辽宁大连116001

出　　处：《应用科学学报》2024年第1期15-26,共12页Journal of Applied Sciences

基　　金：国家自然科学基金青年科学基金(No.62002038)资助。

摘　　要：房颤是一种常见的心律失常疾病,针对现有研究工作大多依赖于单尺度信号段而忽略了不同尺度下潜在的互补信息和数据不平衡问题导致诊断性能下降的问题,提出了一种新颖的基于生成对抗网络(generative adversarial network, GAN)和多尺度残差网络(multiscale residual net, MS-ResNet)的房颤自动检测模型,该网络使用GAN合成具有高形态相似性的单导联心电数据来解决数据的隐私和不平衡问题。同时,设计了MS-ResNet特征提取策略,从不同尺度提取不同大小信号段的特征,从而有效地捕捉P波消失和RR间期不规则特征。该模型联合这两种策略不仅为房颤自动检测生成高质量心电图(electrocardiogram,ECG)数据,还可以利用多尺度网格提取不同波之间的时序特征。在PhysioNet Challenge2017公开ECG数据集上以及平衡后的数据集上评估了MS-ResNet的性能,并将其与现有的房颤分类模型进行了比较。实验结果表明,MS-ResNet在平衡后的数据集上平均F1值和精确率分别达到0.914 1和91.56%,与不平衡数据集相比,F1提高了4.5%,精确率提高了3.5%。Atrialfibrillation(AF)is a common cardiac arrhythmia.However,existing research often relies on single-scale signal segments and overlooks potential complementary information at different scales as well as data imbalance issues,leading to decreased diag-nostic performance.This paper proposes a novel AF automatic detection model based on generative adversarial network(GAN)and residual multi-scale network.The model utilizes GAN to synthesize single-lead ECG data with high morphological similarity,hence address-ing data privacy and imbalance issues.A multi-scale residual network(MS-ResNet)feature extraction strategy was designed to extract the features of signal segments of different sizes from various scales,so as to effectively capture the features of P wave disappearance and RR interval irregularity.The model combines these two strategies not only to generate high-quality ECG(electrocardiogram)data for the automatic AF detection but also to ex-tract temporal features between different waves using multi-scale grids.The performance of MS-ResNet is evaluated on the PhysioNet Challenge 2017 public ECG dataset and a balanced dataset,comparing it with other existing atrialfibrillation classification models.Experimental results show that the average F1 value and accuracy rate of MS-ResNet on the balanced dataset are 0.9141 and 91.56%,respectively.Compared with the unbalanced dataset,F1 increases by 4.5%,and the accuracy rate increases by 3.5%.

关 键 词：心电图 房颤 生成对抗网络 多尺度 自动检测 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]