Application Research On Ultrasonic Blood Flow Doppler Estimation Using Kalman Filter

Ultrasonic Doppler detection is a real-time and noninvasive technology for blood flow estimation. It can extract the pathological parameters and evidence for the diagnosis of vascular diseases, or the pathological changes of the blood circulation system. Therefore, ultrasonic Doppler detection is widely used in clinical application. The key processing method in ultrasonic blood flow Doppler estimation is the spectrum estimation algorithm. This thesis mainly focuses on the two conventional applications(blood flow Doppler spectrum imaging and fetal heart rate detection) of the ultrasonic blood flow estimations. However, these applications are still primarily based on the classical spectrum estimation algorithm. Kalman filter, as the optimal linear filter, is commonly used in areas such as radar, communication, navigation, voice recognition, image processing, and pattern recognition. It has been proved to be the most powerful and effective tool. Kalman filtering estimation is the optimal linear algorithm base on the minimum variance criterion. This thesis applies the Kalman filter algorithm in blood flow Doppler estimation, and concentrates on two conventional applications that mentioned above.In this thesis, overviews of blood flow Doppler spectrum and fetal heart rate detection were investigated, as well as the present used spectrum estimation algorithm. The combination of the Kalman filter and ultrasound Doppler spectrum estimation was studied. The principle of the linear optimal filtering, adaptive cycle theory and the modeling process was stated. For ultrasonic Doppler signal, the complex Kalman filter parameter model was established. Simulation indicated the effectiveness and accuracy of the Kalman filtering modeling, and the advantages compared with the classical spectrum estimation.The Kalman filter algorithm exploration and the experimental verification results in blood flow Doppler spectrum imaging and fetal heart rate detection were discussed separately. The algorithm part stated the parameters modeling and the filtering processing, and the experiment part stated the different experimental platform designs and the experiment results discussions in detailed. For blood flow Doppler spectrum imaging, the complex Kalman filter algorithm achieved the full depth of the blood flow information detection, multimodal Doppler one-dimensional and two-dimensional detection spectrum imaging, realized accurate speed estimation and direction recognition. Based on the consequence of clinical detection imaging, the comparison between Kalman filter algorithm and FFT algorithm was also implemented. In conclusion, Kalman filter algorithm has advantages of flexible parameter design, slight time delay and estimation compensation. For fetal heart rate detection, Kalman filter combined wavelet transform algorithm in parameter modeling. The detection realized accurate heart rate estimation without prior knowledge, envelope extractions, double frequency interference suppression and signal drift control. The calculation cost was negligible so that it has practically productive possibility. Comprehensive experimental results of two blood estimation applications are shown that Kalman filter estimation has excellent estimation accuracy and flexible parameter modeling ability. Ultrasonic Doppler blood flow estimation using Kalman filter can be a successfully combination.