Delicate seperation of Doppler blood flow and vessel wall beat signals by using the EEMD-based algorithm

Based on the ensemble empirical mode decomposition (EEMD) time-frequency analysis for avoiding mode mixing, an algorithm to delicately separate the Doppler blood flow and vessel wall beat signals is proposed in this paper. Firstly, the proper amplitude of added noise and number of ensemble average for noise cancellation are estimated, and then the mixed Doppler ultrasound signal is decomposed into IMFs by using EEMD method. Finally, the IMFs around the division between the blood flow and vessel wall signals are delicately separated using soft-threshold denoising method. Experiments on both computer simulated with WBSR of 20dB, 40dB and 70dB as well as real human carotid Doppler ultrasound signals are carried out to compare the proposed method with the high pass filter, the original empirical mode decomposition (EMD) method and the improved EMD delicate separation method. It is shown that method proposed in this paper provides the highest accuracy of extracting blood flow signals by elimination of the mode mixture, especially for those signals with larger wall-to-blood signal ratio.