Based On Hilbert-huang Power Spectrum To Extract The Blood Vessel Wall Pulsation Displacement Improvement

Cardiac and cerebric vascular diseases, which include cerebrovascular disease and cardiac vascular disease, remain the high rates of suffering, disability and death. The results from recent studies showed that it played an important role to reveal the reasons caused the cardiac and cerebric vascular disease through the detection of unusual vessel wall elasticity. In recent years, an alternative method, the empirical mode decomposition (EMD) with the Hilbert spectrum was used to estimate the artery wall displacement from continuous wave Doppler signals. This method's algorithm is simple, data processing amounts are small, computation time is short and calculation velocity is quick, the experimental results have proven its practicability and feasibility. However, the original EMD algorithm suffers from two fundamental problems: end effects and mode mixing. These are significant drawbacks, which could result in an inaccurate measurement of artery displacement, since when the end effects occurs, the two ends of time series will disperse and the dispersion would empoison in the whole time series gradually, which makes the results get distortion, especially to blood wall displacement signal which has low frequency. Furthermore, when mode mixing occurs, an intrinsic mode function (IMF) will lose its physical meaning by itself, suggesting falsely that there may be different physical processes represented in a mode.In this paper, the EMD with periodic extension (EMD_PE) is proposed for end effects restraint and ensemble EMD (EEMD) is used for mode mixing elimination, in order to improve the HS of the bi-directional Doppler moving wall signals and to get the wall moving displacement more accuracy.The simulation experiments are performed to validate the proposed and used approaches. In the simulation experiments, firstly, 50 quadrature mixed Doppler signals are simulated according to the computer simulation's model of the Doppler signals from bi-directional moving wall and blood flow proposed by Fish et al. Then, each unidirectional signal is decomposed into intrinsic mode functions by using the original EMD, the EMD_PE and the EEMD algorithms, respectively, and the relevant IMFs that contribute to the vessel wall components are identified according to the wall-blood signal ratio. Finally, the Hilbert spectrum for each unidirectional signal from the IMFs of vessel wall is estimated, then the maximum velocity waveform is extracted from estimated spectrum, and then the distension waveforms of the vessel wall are obtained by integrating the maximum velocity waveforms. The mean and standard deviation of root-mean-square error between the estimated (by the EMD_PE and the EEMD algorithms) and the theoretical distension waveforms are 4.2±0.6um and 7.8±1.4um, which are smaller than that, 16.2±3.2um, based on the original EMD algorithm. These results indicate that the improved detection approaches are effective, and more accurate than the original approach does. This obtained information of vessel wall elasticity could be helpful for early diagnosis and precaution of the cardiac and cerebric vascular diseases.