# Sixth Lichuang Electric Competition# Intelligent chest compression electric defibrillation comprehensive rescue device

Note: * is required

Please fill it out during the registration stage↓

* 1. Introduction to project functions

Please enter content...

  Cardiac arrest is the most serious sudden death disease. Once the patient develops the disease, he or she will be brain-dead in just 5 minutes. About 550,000 people die from cardiac arrest in China every year. The effective rescue time for cardiac arrest is only 4 minutes . The most effective treatment for sudden cardiac arrest is correct chest compressions and timely and accurate electrical defibrillation.

Cardiac arrest has gradually attracted the attention of the government. In recent years, many public places in our country have gradually been equipped with AEDs (automated external defibrillators). However, the success rate of cardiac arrest treatment in China is still less than 1%, which is far lower than the 30% success rate in developed countries. The reason is that the treatment of cardiac arrest is far from just being equipped with an electric defibrillator. The successful experience of developed countries comes from decades of cultivating national first aid knowledge and is difficult to apply to our country's national conditions. The contradiction between the time urgency of cardiac arrest treatment and the scarcity of emergency equipment and professionals has become a major problem in the current cardiac arrest treatment in China.

The emergency doctors we cooperate with have questioned the   existing AEDs: existing AEDs and other emergency equipment have major flaws! The main flaws are as follows:

(1) Chest compression carries huge risks . It is extremely irresponsible to hand over chest compressions to someone without professional training . Wrong operation can be counterproductive and bear legal risks.

(2) The existing equipment is not easy to use . Existing defibrillation equipment often only has defibrillation function and is highly dangerous. Even our R&D personnel dare not use AED easily. During first aid, manual chest compressions and electric defibrillation need to be switched back and forth, which may injure the rescuer.

(3) ECG analysis is not intelligent . Existing equipment is limited to only carrying traditional algorithms, and the accuracy of ECG diagnosis is low, which greatly limits the success rate of treatment.

  In order to solve the current difficulties faced by cardiac arrest treatment in our country and the shortcomings of existing equipment raised by doctors. As shown in Figure 1 , we designed an "intelligent chest compression and electric defibrillator all-in-one device".

Figure 1 Intelligent chest compression and defibrillation integrated device 

  This device is an integrated first aid device that does not require professional intervention . It integrates all the functions required for cardiac arrest first aid. Low risk, full functions, low cost, high intelligence, easy to use, easy to store and maintain . It can effectively solve the problems of high risk, cumbersome use, high cost, low intelligence and difficult maintenance of existing equipment .

  This equipment independently and innovatively designed an rescue equipment that integrates a foldable ambulance stretcher, automatic ECG monitor, automatic ECG analyzer, full-chest automatic chest compression machine, and biphasic defibrillator. A huge improvement in "1+1>2 " .

  The electric defibrillation circuit design of this device adopts a multi-isolation scheme . Compared with traditional non-isolated electric defibrillation circuits, it has the advantages of more stable operation and stronger anti-interference ability. At the same time, thanks to the adoption of a pure digital control scheme , it has the advantage of high control freedom.

  This device integrates electric defibrillation and chest compression in its mechanical structure, which can save more than 30 seconds of rescue time. This is of great significance in the golden 4 minutes.

  This device innovatively adopts the simultaneous transfer and full-function treatment solution, which is our first initiative. As shown in the animation 2 , it further improves the treatment efficiency.

Animation 2 Simultaneous transport and treatment (double-click to play) 

2. Difficulties of the work

(1) This equipment independently developed and designed a special electric defibrillation circuit board based on national standards and complete machine requirements , as shown in Figure 3:

Figure 3 Self-designed special electric defibrillation circuit board 

  We also conducted various tests on the electric defibrillation circuit, as shown in Figure 4 and Figure 5. We used double eggs with an impedance close to the human chest as the experimental test load.

  First, test the bioimpedance. The concave part of the lower waveform is the test waveform:

High voltage charging follows:

Animation 4 Quick charging 1KV (235J) energy 

Discharge after full charge:

Animation 5: Release 100J defibrillation energy based on bioimpedance 

The picture below is the waveform diagram of releasing 100J energy:

The picture below is the waveform diagram of releasing 200J energy:

(2) This design independently designed the chest compression power drive circuit, chest compression control circuit, ECG collection circuit, etc., and integrated them on the main board of the whole machine, as shown in Figure 6:

Figure 6 The complete motherboard

(3) This design independently developed and designed a set of automatic diagnosis programs, with a recognition rate far higher than that of commercial products, reaching over 99.7%. It is also supported by data from emergency departments of cooperative hospitals.

Figure 7 Confusion matrix of ECG signal classification results

(4) This design uses wavelet threshold denoising for ECG signal preprocessing. Compared with traditional algorithms, it has stronger performance and cleaner spectrum, which is beneficial to improving classification accuracy, as shown in Figure 8.

Figure 8 Wavelet denoising results 

(5) This design follows national standards . For example, when defibrillation is not being performed or when the equipment accidentally loses power, it can automatically and quickly discharge the high-voltage electricity in the capacitor, and resume normal operation immediately after defibrillation, etc., in compliance with National safety standards.

  The market potential of this product is huge. The United States is equipped with more than 350 AEDs for every 100,000 people, not including other rescue equipment, and China will reach this scale sooner or later. This year Beijing has required every school to be equipped with an AED.

  Due to the government's current support for cardiac arrest treatment and the advantages of its own products, the potential market in China alone is as high as hundreds of billions, and large-scale deployment can save the government hundreds of billions of dollars.

  This product effectively solves the problem of equipment fragmentation during cardiac arrest first aid, effectively solves the harm of AED to the personal safety of rescuers, and effectively solves the conflict between first aid and transport. It has great practical value. The prototype of this product has been developed. , there are currently a number of large-scale medical device manufacturers that have expressed intention to cooperate and have completed preliminary negotiations.

注：主题不限，可以是解决生活/工作中的某个问题、为某个人群/场景设计的方案、毕业设计/课程设计/DIY项目/纯属炫酷项目等。主要讲一下自己通过什么手段解决了什么问题。

*2、项目属性

请输入内容…

本项目为首次公开。

本作品所有软硬件均由队员分工协作原创设计，主板搭载的心电采集电路参考了上届大赛冠军作品《USB便携式ECG心电图监护自动分析仪》并针对本产品应用场景加以改进。

注：请说明项目是否首次公开；项目是否为原创；项目是否曾经在其他比赛中获奖，若有获奖则叙述获奖详情；项目是否在学校参加过答辩。

* 3、开源协议

开源协议：GPL3.0

注：我们允许任何人使用本项目所有的开源内容，但必须注明出处。

注：利他即利己，请认真阅读下述内容。

1. 拥抱开源，赋予项目无限价值。建议项目核心功能开源80%以上；
2. 若某一部分功能不可替代且删掉之后项目无法解决对应的问题，则这一部分实现的功能就是项目的核心功能；比如设计了一台电子负载且设计了一款上位机软件监控功率变化，则电子负载为核心功能，上位机软件为辅助功能；比如电子负载中使用了一款隔离485模块与上位机通信，则此485模块实现的通讯功能为辅助功能；
3. 项目应选择适合自己的开源协议，若项目引用其他开源项目，应注明来源并遵循原作者的开源协议规定；原创项目推荐使用GPL3.0开源协议；
4. 直接引用开源项目的原电路或原代码实现的功能不可作为自己项目的核心功能、使用市场上通用模块直接实现的功能不可作为自己项目的核心功能。

请在竞赛阶段填写 ↓

*4、硬件部分

由于项目原理图较多，在此只展示部分，全部PCB均可在附件下载。

（1）主控单元电路PCB部分原理图（见附件）

（2）电除颤电路PCB（见附件）

（3）电源管理电路（见附件）

      电除颤电路用于产生电除颤的“双相指数截尾”波，该电路会产生用于测量人体的胸阻抗的恒流源脉冲，并通过胸阻抗，和放电电流，以及放电电压实时计算出放电的能量值上传至整机控制电路。

     生理信号采集电路部分负责采集心电数据和胸阻抗数据，实时对心电数据进行分类与分析，计算心率、心熵值等重要数据，判断病人的心脏工作状态并将诊断数据上传至控制电路部分。

     电源管理部分则负责电池的充放电和电量管理，并向电路部分提供合适的电源。该电路需要针对医疗器械的特殊性，针对电源的 EMI 问题进行优化。整机控制电路部分主要负责提供各部件工作的触发信号，并实时接收各部件上传的数据进行分析处理，并进行相应的响应操作。

      电机驱动电路部分主要用于驱动胸外按压机械以及其他各部件的电动机工作，达到控制电机的转速和功率的目的，该设备从控制电路接收触发信号，并反馈电流，电压等电机工作数据。运动传感电路部分则用于感知胸外按压设备的实时运动状态，将状态上传到整机控制电路部分。

 

      整机控制电路部分电路包含主控电路、接口电路、电源电路和模拟电路四大部分，其大致结构图如图所示：

    电除颤电路部分是本设计的电路设计中最重要的一部分，该电路组成如图所示：

   电除颤电路主要包括五部分：

（1）高压储能电容阵列部分，该部分电路用于暂时存储高压放电需要的能量。

（2） 高压电容充电电路部分，该部分电路用于在 20S 内给高压电容阵列充电到 1KV-2KV 左右。

（3）控制电路部分，该部分电路由高性能数字电源专用单片机以及若干外围电路组成。

（4）波形产生电路，该部分电路用于产生高压双相指数截尾波并通过电极片输出到人体。

（5）胸阻抗测量信号调制解调器电路部分，该部分电路用于产生恒流调整脉冲通过电极片输出给人体，并进行解调，得到人体的胸阻抗信号，通过将电极两端信号输出给生理信号采集电路。

本设计会自动采集人体生理信号，并自动分析给予正确的反馈动作，因此本设计需要设计生理信号采集电路。生理信号采集电路主要由信号调理电路、CPU主控电路、通讯电路、电源电路四部分组成，如图所示：

下图为核心电路装机图（整机包含采购的卡片电脑、显示屏等部件，包括但不限于图示电路，但图示电路可独立工作）：

注：推荐使用立创EDA。若选择其他EDA工具，请在附件上传PDF格式的原理图，PDF格式的PCB图纸，Gerber格式的PCB文件。这里可以详细说明您的项目实现原理和机制、注意事项、调试方法、测试方法等。推荐图文并茂的形式向别人介绍您的想法。

*5、软件部分

（1）主控单元固件（见附件）

（2）电除颤单元固件程序（见附件）

（3）电源管理单元固件程序（见附件）

（4）卡片电脑控制界面程序（见附件）

（5）自动分析算法程序（见附件）

由于篇幅限制，在此只介绍我们的最核心的软件，即心电自动分析程序，其它所有源码均可在附件下载！

    心电信号自动识别算法流程如图所示，主要对心电信号进行预处理、特征提取、信号分类等操作，

    由于心电信号通常十分微弱，采集过程中容易受到各种因素的干扰，心电信号的噪声种类繁多，所以心电图分类识别的首要步骤就是对信号进行预处理。国内外研究学者针对去噪方面做了很多研究，主要的去噪手段有经典的数字滤波器和基于小波变换的阈值去噪等。④经典的数字滤波器根据频率范围的不同对噪声进行去噪：对于基线漂移使用高通滤波器去噪、对于肌电干扰使用低通滤波器去噪、对于工频干扰使用带通滤波器去噪。近年来，小波变换技术的快速发展催生出了一系列基于小波阈值去噪技术。该类技术是根据信号和噪声的频率在不同尺度上的分布，先对信号进行小波变换，再根据阈值对各层小波系数进行处理，最后重构信号实现去噪。小波阈值去噪技术对于非平稳信号具有优秀的处理效果，与传统处理方法相比有显著的优越性。

    On the basis of preprocessing and feature extraction of ECG signals, deep learning is used to automatically classify and identify ECG signals. The research on deep learning is inextricably linked to neural networks. In 2006, Hinton proposed the deep belief network, which became a milestone in the field of deep learning. Nowadays, in the context of big data and computer computing power increasing significantly, deep learning has made great progress in research in many fields such as image processing, speech recognition, and NLP. The convolutional neural network model has experienced continuous evolution from the earliest LeNet model to the current GoogleNet, ResNet, etc., and the recognition rate is already very good. When studying ECG classification algorithms, the advantages of deep learning in processing large amounts of data have significantly improved the classification effect. The architecture of the ECG signal classification model is shown in the figure. It mainly performs wavelet transformation on the input signal and then

Perform wavelet threshold denoising, and then perform inverse wavelet transform on the denoised signal to obtain the denoised signal. The classification model adopts CNN architecture as a whole. One-dimensional convolution can effectively extract ECG signal features, and then send the extracted ECG signal features to the fully connected layer for further feature extraction and training. Experiments show that this model architecture has a good impact on ECG signals. The classification effect is significant.

The figure below shows the confusion matrix predicted by the classification model. For 21543 normal ECG signals, the number of correct predictions is 21513, for 513 atrial premature beat data, the number of correct predictions is 518, and for 2124 ventricular fibrillation data, the number of correct predictions is 21513.

The number of correct predictions was 2106, 1914 for left bundle branch block pulse data and 1516 for right bundle branch block pulse data, and the number of correct predictions was 1912 and 1513 respectively. It can be seen that the model has a very good classification and prediction effect on ECG data.

The figure below shows the ROC curve and AUC value predicted by the model. It can be seen that the model's predicted AUC values ​​for the five categories are close to 1, which further illustrates that the model's classification of ECG signals is very accurate.

 

Note: If your project involves software development, please upload the corresponding project source code in the attachment. Here you can describe in detail your software flow chart, functional module block diagram, explanation or popular science of related algorithms, source code structure, construction and configuration of compilation environment, source code compilation method, program burning method, etc. It is recommended to introduce your ideas to others in the form of pictures and texts.

*6. BOM list

The following is the BOM of the main control PCB, see the attachment for details:

The following is the BOM of the defibrillation unit:

The following is the power management unit BOM:

Note: BOM list involved in the project. Please upload a screenshot of the BOM at this location. Please upload the list details in PDF format to the attachment. Suggestions include model, brand, name, packaging, procurement channels, usage, etc. The specific content and form should be based on clearly expressing the project composition.

*7. Contest LOGO verification

Please upload a project picture containing the competition logo. The logo will be printed on the PCB in the form of silk screen printing.

Click the zip to download the competition logo! (Contest logo).zip

* 8. Demonstrate your project and record it as a video for uploading

Video requirements: Please shoot horizontally, with a resolution of no less than 1280×720, in Mp4/Mov format, and the size of a single video is limited to 100M;

Video title: Lichuang Electric Competition: {Project Name}-{Video Module Name}; such as Lichuang Electric Competition: "Autonomous Driving" - Team Introduction.

More details: https://diy.szlcsc.com/posts/06c94d90c2c447dfbd9ed7339ff4a5b1