Studies For Analysis Of Dynamic Electrocardiogram

The guardianship and detection of Electrocardiogram, blood pressure, oxygen saturation, body temperature for people with chronic diseases and old men in the clinic, it has great clinical significance in lone-termly guarding these physiological signal, predicting people's healthy trends and the disease's affection to human bodies. Analysis and identification have been one research focus in current signal processing field and biomedical engineering.In this article, we use the theory of wavelet transform and mathematical morphology, to study the filtering pretreatment, waveform detection and location. The mainly work are as follow:Firstly, we analyze some data of the current ECG pretreatment technology, waveform detection and location. Compared the advantage and disadvantage of these methods, we find that there are some place which are in need of refinement in current ECG pretreatment technology, waveform detection and location, and the extraction and selection of feature information. Thus, we use the denoising and feature detection algorithm as the topic we study in the article.Secondly, on the basis of studying generation mechanism and feature of ECG physiological parameters, we discuss the frequency, EMG and baseline drift's interference to ECG signal. Although the wavelet filter has the advantage of time-frequency localization, it causes distortion in filtering low-frequency noise in the P waveform and T waveform, the processing time is long, real-time index is not good. So, using wavelet transform and morphology theory, the article projects a wavelet threshold denoising algorithm combining wavelet transform and morphology theory, removing the noise efficaciously, reserving the singularity of ECG signal, promoting the signal to noise ratio effectively. We do some emulational experiments for this algorithm using the MIT-BIH ECG database, after the experiment, we found that all of the MSE, SNR and LocalMax are promoted. So this experiment indicates the algorithm raise the noise SNR effectively, and the denoising effect is also very good.Thirdly, we analyze the boundedness of traditional wavelet algorithm in waveform detection when dealing with the problem of undetected signal. On this basis, the article improves the algorithm based on wavelet transform's waveform detection and feature identification, projects an improved algorithm based on quadratic B-spline wavelets used in QRS waveforms detection. It decomposes the ECG signal to different frequency spectrum, adds re-examination and compliment tactics, so that the undetected and miscarriage of justice will be avoided. The algorithm gets a good result. In the end, emulational experiments are done to testify the effectiveness of processing ECG signal with noise, and it does well in the total false detection count, the total missed count and missed rate.At last, we summarize the main work of this article, and look into the future of the next work we can do.