Multifractal Detrended Flucturation And Hurst Index Based Analysis Of The Pathological Electrocardiogram Signals

The analysis and diagnosis of electrocardiogram(ECG) signals is one of the hot researches insignal processing field.Effective identification and diagnosis of heart disease will promote themedical industry and raise the level of people's life.Multifractalily is a very active new branch of thenon-linear science,it has now been widely used in various fields as a major direction of the fractalgeometry field.In this thesis, multifractal detrended fluctuation analysis method is used to analyze themultifractal characteristics of the sinus bradycardia(SBR)signals, the atrial fibrillation(AF) signalsand the normal electrocardiograph (ECG) signals. It is aimed to be helpful with clinical diagnosis.By analyzing the Hurst index, Renyi index and multifractal spectrum, it is found that when the orderof fluctuation function is positive, the three kinds of signals show distinct long-range correlationproperties. To compare with the three signals' multifractal spectrum, the width of the normal ECGsignals' spectrum is smallest, the SBR signals' is of the second stronger fractality and the AF signals'is of the biggest.Finally, this thesis uses the rescaled range(R/S)analysis to distinguish between the normal ,congestive heart failure and sudden cardiac death signals with the Hurst index.The results show thatall the Hurst index of the three signals are larger than 0.5,but the values are vary. It indicates that thelong-term relevant character is the best for normal signals, better for congestive heart failure ,andthe worst for sudden cardiac death.So Hurst index can be used as an identification criterion fordistinguishing them.The study provides good reference for clinical diagnosing and distinguishingthe pathological electrocardiogram signals from normal signals.