Estimation Of Central Aortic Pulse Wave Based On Multi-channel Blind System Identification

Nowadays, pulse wave is very important in predicting and treatment of Cardiovascular System (CVS) diseases because it provides tremendous physiological and pathological information of human CVS. So far, blood pressure measured at brachial artery is used as a routine method in diagnosis of CVS diseases; however, as the afterload of ventricular ejection, Central Aortic Pulse wave (CAP) is generally more informative about cardiac dynamics and global circulation. In order to monitor CVS noninvasively and effectively, a new method based on Multi-channel Blind System Identification (MBSI) was proposed in this thesis to reconstruct CAP from one or two channel noninvasive Peripheral Aortic Pulse wave (PAP).Background, significance and relative achievements of former researches were introduced in the introduction of the thesis. In chapter2, physiological properities and modeling approach of CVS were described in detail, especially development and application trends of the Windkessel model of CVS. A "gray box" model of CVS was designed based on Windkessel theory and two-channel "T-tube" model in chapter4, the model provided the theory foundation for MBSI algorithm.In chapter4, the equivalence of the IIR and high order FIR function has been proved. The conversion from "gray box" model to "black box" model greatly simplifies MBSI algorithm. Effectiveness of MBSI algorithm can be verified by three kinds of data:whether MBSI algorithm will be affected by change of CVS status can be verified by data from "T-tube" model; The best PAP measure site can be selected by CAP and PAP recorded from animal experimental, furthermore, MBSI algorithm can be compared with other noninvasive CAP measurement method (eg:the transfer function method); Clinical data was used to verify the improved MBSI algorithm-SBSI algorithm and evaluate its practicality.Verification results from the first two kinds of data show that MBSI algorithm can estimate CAP stably and accurately. Overall error rate, error of systolic pressures and error of diastolic pressures produce by MBSI algorithm are5.1%,2.8mmHg,2.2mmHg respectively, compare with transfer function method, the three types of errors improved by3.7%,2.8mmHg,1.7mmHg. It was proved by12clinical data that error of systolic pressures and error of diastolic pressures produce by SBSI algorithm are very small (3.2mmHg,2.4mmHg). SBSI algorithm is promising to be used in noninvasive CVS monitoring.