Design And Implementation Of Synchronous 12-lead ECG Comprehensive Analysis Software

In recent years, as the number of patients with cardiovascular disease and elderly population is growing, the tension of medical resources is gradually emerging. In order to alleviate this situation, the embedded system and technology laboratory of Tsinghua University develops the remote dynamic real-time ECG system. This system can collect and analyze real-time ECG data, remind ECG abnormalities, carry out a comprehensive analysis of the previous ECG data which is stored in the database, and generate a diagnostic report. It not only improves medical level and efficiency, reduces health care costs and the occupancy of medical resources, but also meets the health needs of the people.Synchronous 12-lead ECG comprehensive analysis software, which is also called ECG comprehensive analysis system, is an important subsystem of remote dynamic real-time ECG system. It can download the previous ECG data to the local computer from the database, carry out an automatic analysis and discriminate arrhythmias, count the results, and generate a diagnostic report to feedback to the doctor and the patient.This paper describes the design and implementation of synchronous 12-lead ECG comprehensive analysis software which is based on support vector machine. Firstly, the author elaborates the overall architecture of the software, then researches and improves the pretreatment of ECG signal, ECG waveform detection algorithm and arrhythmias discrimination algorithm, and finally describes the implementation of each function module of the software. The optimization of arrhythmias discrimination algorithm, and the design and implementation of the software are the focuses of this paper.The main research achievements of the paper are as follows:1. A zero-phase filtering method is used to optimize the classical differential filter, and an improved filtering algorithm is proposed to preprocess the ECG signal, then the defect which is phase shift has been solved.2. An RR interval prediction mechanism is introduced to optimize the ECG waveform detection algorithm which is based on threshold, and an improved ECG waveform detection algorithm is proposed. It not only improves the accuracy of the R-wave detection, but also speeds up the processing speed.3. A new method based on multi-lead features is proposed, it will extract features from all leads, and the feature difference of different arrhythmia types has been strengthened compared to the single-lead features. When we use SVM to analyze ECG signal and discriminate arrhythmia types, the classification accuracy has been improved.4. An overall architecture of the ECG comprehensive analysis system has been designed, and the improved algorithm has been applied to the system, and the synchronous 12-lead ECG comprehensive analysis software has been programmed. Then we improve the content of diagnostic report in accordance with Chinese national standards, and the software can be applied in clinical practice.