An ICA Based Estimation Of Ambulatory Blood Pressure From The Radial Pulse Waveform

Continuous ambulatory blood pressure(ABP) records may be an important independent determinant to the better prevention and diagnosis of cardiovascular disease.In order to find a noninvasive, continuous and convenient blood pressure(BP) measuring method, this thesis studies the correlation between the radial artery pressure waveform(RAPWF) and BP, the interference factors in the process of RAPWF acquisition and the forecasting model to derive BP from RAPWF.The ABP can be calculated by the characteristic parameters of RAPWF.Thesis verified the correlation between BP and RAPWF, and found that the interference factors exist in the process of RAPWF acquisition will affect the characteristic parameters.To test which factor will affect the characteristic parameters, we design the factor analysis experiment selecting the BP change, tightness of wrist strap, the location of sensor and the different position this four factors. The experimental results show that not only the factors influence on RAPWF respectively, and there is interaction between factors. To eliminate the effect and the correlativity among measurement factors, independent component analysis(ICA) method is used to decompose the raw signal. Based on the spectrum density of independent component, an auto-regression forecasting model is set up to derive BP. An experiment to validate the prediction model is done and the results show an excellent correlation and agreement between directly measured BP which is measured by Omron electronic BP monitor(Type: HEM-7012). When the ICA-ARF is used, the error reduces from 7.4mm Hg to 2.55 mmHg. The novel methods have important implications for the simplification of noninvasive CASP measurement and 24-hour ambulatory BP monitoring and assessment of central aortic blood pressure.