The Study On Holter Analysis System

Dynamic Electrocardiography(DCG)can continuously record electrocardiosignal (ECG) for 24 hours or longer, while the patients can active normally. With the help of modern computer, physicians can analysis the signal, find out all kinds of arrhythmia events and ST segment abnormal events, thus they can get important foundations for diagnosis. Holter system is widely used in cardiovascular diseases, it is a practical, efficient, noninvasive, safe, accurate and reproductive method for clinical diagnosis and other medical studies.On the basis of investigation on the existing Holter system and software at home and abroad, I put forward better algorithms and realizing programs, so that the diagnosis is more accurate and make up the shortcomings of currently used analysis software. The main works in this paper are as follows:First,the preprocessing of ECG signal-filtering is carried on. The interference includes 50Hz and its higher harmonics, muscle artifact and baseline wander and so on. After compared lots of literatures and methods, we offer the norm for selecting different filters. That is to say, in arrhythmia analysis, the integer coefficient lowpass filter is adopted, while in heart rate variability analysis, the high pass filter designed by all-pass minus low-pass filter or filter which can remove both baseline wander and 50Hz power-line frequency at one time is used. Second, template establishment and classification of clinical ECG are accomplished. The key points of ECG wave are accurately marked, as for the nonstandard waves, special methods are used. Detection method of ST segment is discussed in detail and atrial fibrillation detection is explored in this chapter as well. Third, analysis on heart rate variability signal is accomplished. Because HRV signal is unevenly sampled, direct spectrum estimation will cause wrong result. Resample on HRV signal is necessary. In this chapter after comparing different evenly sampling methods, the best way -cubic spline interpolation is adopted in spectrum estimation. Then HRV signal is analyzed in time domain, frequency domain and chaos as well. Several questions in classical spectrum estimation are presented.At last the extrapolation of HRV signal is obtained. Since the detected signal isalways limited, according to the uncertainty principle, the frequency band is infinite. In order to get accurate spectrum estimation result, extrapolation is necessary. In this paper the extrapolation object is the shorten polynomial but not the original signal itself. This algorithm is much easier than prolate spherical wave function expansion method.