Emd-based Doppler Ultrasound Blood Flow And Wall Signal Separation Method

The Doppler ultrasound, which can measure the velocity of blood flow noninvasively, has become one of the important means in diagnosing vascular diseases. However, the clinical Doppler ultrasound blood flow signal is usually corrupted by vessel wall components which would interfere with the measurement of the blood flow velocity, and consequently influence the accuracy of diagnose. So it is of great significance to precisely extract the intact blood flow components from the mixed signal.In this dissertation, two novel methods are proposed to separate the blood flow and vessel wall signals, namely the EMD separation algorithm with an adaptive wavelet threshold and the EMD separation algorithm method based on the instantaneous frequency. Both have fully exploited the characteristics of the blood flow and wall clutter in time and frequency domain. These two methods are applicable to not only unidirectional signals but also bidirectional signals.To test these two methods, unidirectional and bidirectional blood flow signals are simulated for both the continuous and pulsed Doppler ultrasound system, and the result of the separation employing our two methods are compared with those using the high-pass filter, the spatially selective noise filtration and the original EMD method. It turned out that compared with other approaches, two algorithms presented in this dissertation boast higher accuracy and stability, whose mean separation error is at least 50% lower than that of the high-pass filter. In the condition that the highest frequency of wall signals are 100 Hz, the EMD separation algorithm with an adaptive wavelet threshold performs better, while the highest frequency of vessel wall are 150 Hz and 200 Hz, the EMD separation algorithm method based on the instantaneous frequency is more suitable.New methods are further applied to clinical Doppler ultrasound signal collected from common carotid artery, and the result also demonstrate that while eliminating the wall components, our two methods could reserve more low-frequency blood flow components than other methods, which is expected in clinical treatment.