The Algorithms Research Of ECG Signals Intelligent Detection & Analysis For Mobile Cardiac Telemonitoring System

Based on mobile communication networks, the novel ECG telemonitoring system can provides special cardiac care for the patients at home or in community. According to the results of real time detection & off-line diagnosis for ambulatory ECG signals transmited into computer through mobile communication networks , it can provides emergency treatment, medical forecast, consultation and instruction to the users with heart disease to ensure the effect of ECG monitoring and to meet the needs of patients with paroxysmal and dangerous heart attack. It is vital to develop detection and diagnosis algorithms for ambulatory ECG signals. At the same time, the algorithms research space of feature extraction and classification for ECG signals exist at present.Based on the demands of academic research and application, this dissertation carry on the algorithms study on real time detection, feature extraction and classification of ambulatory ECG signal for the mobile cadiac telmonitoring system. It mainly includes the following distinctive works accomplished: (1) A precise real-time QRS detection algorithm based on local minima of second derivatives and cross-zero points of first derivatives of ambulatory ECG signals to meet the need of mobile ECG telemonitoring system is developed in this dissertation. The novel algorithm filters power-line interference and most of muscle noise and is insensitive to baseline draft and noise caused by mobile communication. The QRS recognition thresholds, which can revise themselves according to the detected values and vary with the analyzing signals, are designed. For the part of data of MIT/BIH arrhythmia database, the simulating ambulatory ECG signals and clinical ECG signals,this algorithm correctly detects up 99.12%, 99.92% and 99.97%. This novel method improves the signal-noise rate of second degree difference signals and first degree difference signals and has more advantages in feature point's test by the way of linear processing. Simultaneously as a result of setting threshold with the function of self-adaptation and self-study , it improves the accuracy of the detection and the orientation of QRS complex and also reduces the time searching window not more than 0. 2 second . It shows the algorithm is feasible.(2) Feature information of ECG are extracted using rough set theory firstly and creatively in this dissertation due to classical feature information of ECG signals are extract by experience and subjectivity of researchers. Decision-making table of ECG is established by knowledge represent system of rough set theory with all coefficients that could affect ECG signals disease type as a condition attribute and all diagnose results of ECG as decision attribute of ECG knowledge expressing system. Then indiscernibility relation and knowledge reduction of rough set theory is applied to remove redundant condition attribute. It can conclude that the kernel condition attribute is the feature information of ECG signal. This feature extraction method is processed in term of mathematics and provides an academic foundation to feature extraction for automatic analysis of ECG signals.(3) The multi-domain feature information of ECG signals is analyzed and fused by rough set theory in the dissertation. With a view of data mining & data fusion, it is brought forward that the feature information of ECG monitoring should be composed of static feature information of the patient, time-domain feature subset of ECG signal, frequency-domain feature subset and wavelet-domain feature subset of ECG signal. In this paper rough set theory is applied to extract feature in time-domain, frequency-domain and wavelet-domain information of ECG signal concretely. Various multi-domain feature information is introduced as ECG feature information finally. As a result the defect of feature extraction that separately depend on the information of time-domain, frequency-domain and wavelet-domain is overcome.(4) A classification algorithm of cardiac arrhythmia by support vector machine is studied in this disserta...