Extract Ecg Fractal Feature Vector Algorithm Based On Nonlinear Dynamics

No matter the heart is normal or arrhythmia, recent studies show that it is a chaos system. However, it is hard to make accurate judgments for non-linear signals if we choice traditional time or frequency measures. Fortunately, we find the quantity and quality of the system can be analyzed exactly by the theory of non-linear.As all the physiological information is contained by a single time sequence, we apply the time sequence to the theory of space-reconstruct in order to analyze and study the dynamic characters of the heart. In this study, we investigated measures of fractal and chaos theory in regard to pathological diagnoses of ECG as well as pick-up correlation dimension of ECG. An improved algorithm is brought forward for the correlation dimension estimation in an automatic way, which can overcome a series of shortcomings, such as subjectively choosing correlative parameters, inaccurately computing, choosing scaleless range through experimenters' own experiences and observation. An improved algorithm is brought forward for the correlation dimension estimation in an automatic way, which can overcome a series of shortcomings, such as subjectively choosing correlative parameters, inaccurately computing, choosing scaleless range through experimenters' own experiences and observation. It has been shown that the improved method, which introduced the GAs algorithm, can determine the scaless range with more precision, more objectivity and higher linear. As an experiment example, this algorithm has been applied to 30 groups ECG data of MIT-BIH Arrhythmia Database. Analysis result demonstrates that the approach is able to obtain fractal characteristic vectors of ECG in an automatic way.