| 摘 要: 针对可穿戴设备的长时间心电记录、实时分类及对心电数据的远程监测分析问题,开发了一个对接医疗级心电采集终端，并实现实时监测、实时分析，并通过深度学习模型自动对心律失常分类的通用系统。该系统中部署的深度学习模型是基于残差网络构建的，深度学习模型的训练和测试使用2017 年心脏病学挑战赛(CinC2017)提供的数据集。训练和测试结果显示，模型具有较好的性能。系统通过反向代理服务器(Nginx)部署在阿里云服务器上，能够稳定运行；心电采集终端贴在患者身上，通过用户App端和医生后端实时反馈系统自动监测分析的数据，并且有较好的分类效果。该系统可用于有心血管疾病风险的人群，起到早发现、早预防的作用。 |
| 关键词: 深度学习；心律失常；Django；云服务器 |
|
中图分类号: TP315
文献标识码: A
|
| 基金项目: 浙江省科技厅重点研发项目(2020C03060)；国家自然科学基金(61672466)；浙江省自然科学基金-数理医学学会联合基金重点项目(LSZ19F010001). |
|
| Design of Arrhythmia Classification System based on Deep Learning |
|
LÜ Hang1, LI Yang1, ZHANG Jucheng2, WANG Zhikang2, JIANG Mingfeng1
|
( 1. School of Computer Science and Technology, Zhejiang Sci -Tech University, Hangzhou 310018, China;
2.The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou 310009, China)
hanker.lv@foxmail.com; dr.yangli@outlook.com; jucheng@zju.edu.cn; akang511@126.com; m.jiang@zstu.edu.cn
|
| Abstract: Aiming at the problems of wearable ECG devices, such as long-term ECG (electrocardiogram) recording, real-time classification, and remote monitoring and analysis of ECG data, this paper proposes to develop a general system of automatic classification of arrhythmia bases on deep learning model. This system docks medical-grade ECG acquisition terminal, realizing real-time monitoring and analysis. The deep learning model deployed in the system is constructed based on residual network. Its training and testing are performed by using the data set provided by 2017 PhysioNet/Computing in Cardiology Challenge (CinC2017), and the training and testing results show that the model achieves good performance. The system is deployed on the Ali Cloud server through Nginx (A reverse proxy server) and has been able to run stably. The ECG acquisition terminal, which is attached to the patient, automatically monitors and analyzes the data through real-time feedback system of the user App terminal and the doctor back-end, and it shows a good classification effect. The proposed system can be used for anyone with a risk of cardiovascular disease, playing the role of early detection and early prevention. |
| Keywords: deep learning; arrhythmia classification; Django; cloud server |
