Research On Photoelectric Volume Pulse Wave Imaging Mechanism And Image Data Mining

At present,all walks of life in our country are developing rapidly and the people's living standard is improving continuously.More and more people are pursuing material and spiritual life satisfaction,but they are also increasingly concerned about personal health problems.However,bad living habits lead to frequent cardiovascular diseases.At the same time,with the aging of the population in our country becoming more and more serious,the prevalence rate of cardiovascular diseases is increasing year by year,and a simple and feasible monitoring scheme of cardiovascular physiological parameters is urgently needed.IPPG(Imaging Photolethysmography,IPPG)technology has the advantages of non-invasive,low-cost,easy to operate and so on.It fits well with the daily physiological monitoring needs of cardiovascular patients in China and has become one of the emerging research hotspots in the field of physiological parameter measurement in recent years.However,IPPG technology is still in the research stage at present,and there are still some defects,such as single measurement parameters and low measurement accuracy.In view of the defects of IPPG,this paper firstly makes a systematic analysis on the optical principle of IPPG technology.On this basis,theoretical modeling and algorithm improvement are carried out.The specific research contents are as follows:(1)The optical and physiological principles of IPPG technology are systematically analyzed.The heart rate,blood oxygen saturation,The physiological significance of blood pressure is introduced,and the measurement principles involved are analyzed and summarized,such as time-frequency domain heart rate value calculation,double optical path blood oxygen saturation measurement,elastic cavity theory,etc.(2)Aiming at the problem that pulse wave is easily interfered by noise,this paper presents a set of relatively complete presets.De-noising algorithms include abrupt noise removal based on differential outlier detection and cubic spline interpolation,and high frequency and baseline drift noise removal based on zero-phase digital band-pass filtering,wavelet transform algorithm and morphological filtering.Aiming at the problem of extracting peak-valley eigenvalue information,this paper improves on the basis of extreme point method.The improved algorithm can adaptively and alternately search peak-valley eigenvalue information,which lays a foundation for subsequent cardiovascular parameter calculation.(3)Considering that there is a certain spectral width of illumination light sources in practical applications,this article cited A multi-color light transmission model is introduced.When calculating the heart rate value,there is the problem of missed detection and false detection in feature point extraction.Before calculation,this paper uses the correction algorithm to correct the number of peak and valley points,thus improving the heart rate measurement accuracy.Considering that the heart rate value error is large when noise interference is serious,this paper uses the K-means clustering algorithm to improve the heart rate measurement accuracy under strong noise.Combined with the multi-color light transmission model,this paper establishes a non-linear blood oxygen saturation measurement model through theoretical derivation.Compared with the linear model,it reduces the monochromaticity requirement of the light source,and at the same time,it can improve the measurement accuracy of blood oxygen saturation to a certain extent.In the calculation of blood pressure,this paper carries out correlation measurement research based on pulse wave waveform,and takes discrete sampling of a single pulse wave period as the input of BP neural network directly,thus avoiding the complicated extraction process of pulse wave characteristic parameters and utilizing the MIMIC number of physionet The results show that the mean absolute error of systolic blood pressure is 6.63 mm Hg and the standard deviation is 8.39 mm Hg;.The average absolute error of diastolic blood pressure is 13.35 mm Hg and the standard deviation is 16.8mm Hg,which proves the feasibility of blood pressure measurement method based on pulse wave waveform to some extent.This paper independently designed a cardiovascular parameter calculation platform based on MATLAB programming language,and Comparative experiments of heart rate,blood oxygen and blood pressure were carried out,and the experimentalresults further verified the reliability of the algorithm.