Emd Adaptive Filtering To Improve The Accuracy Of The Wall Doppler Signal Separation

With the improved quality of life, high morbidity, mortality rate of cardiovascular disease has become a serious threat to human life and disease. The early diagnosis can improve treatment of diseases; the disease has a major role in the treatment process. In the early stage of cardiovascular disease, the velocity of the blood flow which in the middle part of the vessel wall could not be changed obviously; however, the velocity of the low-frequency blood flow which near by the vessel wall could be changed obviously, then ultrasound Doppler signal of the low-frequency blood flow will take a lot of information which is about cardiovascular disease. Therefore, to more accurately detect changes in blood flow near the vessel wall, will contribute to early diagnosis and treatment of the cardiovascular disease. Non-invasive ultrasound Doppler is more effective and widely used in the detection approaches to cardiovascular disease. If we can find an approach which could be detected the changed blood flow accurately, it will be contribute to the early diagnosis of cardiovascular disease and treatment. But the ultrasound Doppler signal of the vessel wall is low-frequency and high amplitude; it's so easy to mix with the blood flow which is low-frequency when we are detecting the blood signal, and so hard to accurately separate the blood flow signal and the vessel wall signal. For this how to separate the vessel wall signal and the blood flow signal, improve the accurate rate of the separation is an important research object.In recent years, there are many approaches to separate the ultrasound Doppler blood signal, the Empirical Mode Decomposition (EMD) is the most objective approach which is the easiest to achieve and the most effective. We can indicate this part of the residual part is vessel wall, but the residual part which mixes with a part of the low-frequency blood signal. In order to detect Doppler signal can be as accurate as possible to retain low-frequency blood components, this thesis is proposed Empirical Mode Decomposition based-adaptive filter (EMD_AF) based on the EMD algorithm. The EMD_AF algorithm combines the EMD algorithm with adaptive filtering algorithm. The principle of the two algorithms is decomposes the mixed blood signal into many IMFs by EMD algorithm, and let these IMFs through adaptive filter, thronging continuous self-tuning the filter to the desired signal (signal related to the vessel wall signal) around, then get the smallest error of the output signal (extracted from the blood vessel wall signal). This separate approach could extract more accurate blood flow signal, and could be more accurate to retain the low-frequency blood flow nearby the vessel wall components. EMD_AF algorithm combines EMD algorithm and adaptive filtering algorithm, and it could be better to handle non-linear, non-stationary signal.To test this new approach, we utilize the computer to simulate the blood vessel wall signal and Doppler blood signals mixed-signal, and to compare new approach with traditional EMD algorithm. The results show that: the new approach could separate vessel wall signal accurately, while acquire most of the low-frequency blood flow components and extract more accurate ultrasound Doppler signal. The performance of the new approach is more superior to the traditional EMD algorithm. It could provide us with more accurate low-frequency blood flow components, and providing accurate important basis to diagnosis of the cardiovascular disease in the early stage.