The Real-time ECG Processing And Detection Technology Based On Cortex

Nowadays, heart disease has been an important factor that threatens human health. Ventricular arrhythmia and sudden cardiac death (SCD) are the sever problem to medical researchers, because of their incidence of a sudden and lack of timely treatment. Therefore developing a portable electrocardiogram monitor system is of great importance to the detection and diagnosis of heart disease. ECG real-time processing and detecting technology is one of hotpots in the current signal processing fields. In this paper, the techniques of ECG preprocessing, validity detection and supraventricular QRS recognization are discussed respectively. All algorithms above are implemented on hardware system platform. STM32F103 series chips with the kernel of cortex-M3 are used to build the hardware system.ECG preprocessing technology research: as the monitoring system actually needs the ECG analysis an diagnosis, this paper proposed a simple synthetic ECG denoising algorithm based on the digital filtering and modern signal processing technology. To deal with these different kinds of noises, three distinct techniques were used. Firstly, baseline wander was removed by the FIR high-pass filter with Kaiser windows. Then, a simple and narrowband power-line interference notch filter was designed to suppress power-line and harmonic interference. Finally, EMG overlapping with the cardiac components in the frequency domain was suppressed by wavelet denoising. The SNR and improved SNR were shown in this paper. This synthetic algorithm is easily implemented by microprocessor in real-time processing and may therefore serves as an effective methods for filtering noisy ECG.ECG validity detection research: Cardiogram recordings include numerous noncardiac containments, which overlap with the cardiac components in the frequency domain. Despite the rich literature in ECG denoising field, there are still noise left in the ECG. Useful information can not be extracted from such ECG. Therefore, the detection ECG validity algorithm was proposed based on wavelet transform. First, signals were preprocessed by suppressing the noise. Then, they were calculated by this method with 4 s slipping windows. In this way, accuracy and efficiency are improved greatly for the subsequent processing such as ventricular fibrillation (VF) detection. The algorithm is successfully evaluated on the complete MIT-BIH Arrhythmia Database, some data from CU Database and the MIT-BIH Noise Stress Test Database. Results show that the algorithm has a high quality and is easily implemented. It may serve in the real-time remote health monitor system.QRS and supraventricular QRS recognization research: the classic defference threshold arithmetic and wavelet transform method were used respectively to detect the QRS wave. Quadratic b-spline wavelet and a-trous algorithm were used to implement the wavelet transform. According to the characteristic parameter, position of R wave and the start and end of QRS can be localized. Furthermore, the width of QRS and instantaneous heart rate also can be computed. This paper gave some simulation result, and the experiment denoted the algorithms accuracy rate were above 99%. Generally distinguishing the supraventricular QRS and ventricular QRS was based on the difference of QRS width. This method always caused erroneous judgement. So we put forward a new method which is based on the theory of lipschitz. Simulation results denoted this method not only making up the shortcomings of original method, but also improving the detection accuracy rate.Single channel portable real-time detector should possess such traits as small volume, easy-taking, stable and reliable and incorporating the embedded system, DSP, wireless mobile communication, low consumption and multitask program technology etc. We chose the STM32F103 series chip with the kernel of cortex-M3. Cortex-M3 is specially designed to meet high performance, low power consumption, real-time application, competitive price in the embedded field. Keil u Vision3 was used as software platform and the software was programmed with C language. The foregoing ECG processing algorithms were transplanted into the hardware system to debug and to realize the real-time signal processing.