| The blood biochemical composition directly reflects the health condition of body.Currently, routine clinical testing methods are almost invasive or minimally invasive.Not only existing the hidden dangers of cross-infection, but also needing reagents forin vitro analysis that is difficult to realize online monitoring. So non-invasivebiochemical monitoring has been one of the hotspots by domestic and foreign scholars.Near-infrared absorption spectroscopy (NIRS) technique is invasive, reagent-free, nopollution and has online monitoring, which has a high research value and broadprospects in the field of non-invasive biochemical analysis. However, near-infrarednon-invasive biochemical analysis technology has not reached level of clinicalbecause of the less biochemical components in blood and serious tissue backgroundinterference.In order to eliminate human tissue background interferences,“Blood VolumeSpectral Subtraction Method” was proposed. By subtracting two blood spectrameasured in a row within an extremely short time period, effective blood spectra canbe acquired. However, a more quickly speed and higher signal-to-noise ratio (SNR) ofpulse wave signals collecting were needed for “Blood Volume Spectral Subtraction Method”, the performance of near-infrared spectroscopy instruments in the marketcan’t meet those requirements. So it is necessary to research a rapid and high SNRmethod for volume pulse wave acquisition, in order to improve the performance ofinstrument.This paper focuses on the acquisition method for fast and high SNR spectra in thearea of near-infrared noninvasive biochemical analysis. The hardware circuit designof fast and high SNR acquisition system for volume pulse wave signal and the noisereduction processing of volume pulse wave were mainly included in this research. Thedetails and main conclusions of this research were as follows:1) Based on the requirement of fast and high SNR acquisition system fornear-infrared non-invasive biochemical analysis, detectors with different workingstyles have different characteristics. The near-infrared acquisition systems for volumepulse waves were designed according to each styles of InGaAs detectors of unit type,multi-array type and multiple discrete type. A preamplifier circuit and data acquisitioncircuit were contained of each system.2) The performances of three near-infrared volume pulse wave acquisitionsystems were tested, and the advantages and disadvantages of three systems wereanalyzed. Then the system which most satisfied the requirements of near-infrarednon-invasive biochemical monitoring was selected. It was proved by relatedexperiments that the SNR of unit style volume pulse wave acquisition system wasabout28000：1, and the dark noise was about20μV. Although the unit style systemhad a higher SNR, the spectrum was obtained only by scanning. Not only theacquisition speed was limited, but also it was very difficult to ensure the accuracy ofspectrum that was due to the blood fluidity.256pixels InGaAs detector G9211wasused in multi-array volume pulse wave acquisition system, which can collect256signals at different wavelength simultaneously with a fast speed. But the average SNRof this system was7000：1, which can’t meet the requirements of non-invasivebiochemical analysis.16pixels InGaAs detector G7150was used in multiple discretevolume pulse wave acquisition system, which can collect16signals at different wavelength simultaneously. About50spectra can be captured per second at most. TheSNR of this system can reach13000：1, and dark noise was about43μV. In summerconsidered, the volume pulse wave acquisition system of16pixels discrete style wasthe best system for near-infrared non-invasive biochemical analysis.3) Clinical experiment was completed by using the volume pulse waveacquisition system of16pixels discrete style. Eighty-one fingertip absorption curvesof different ages and gentles were collected, and concentrations of hematocrit andhemoglobin were predicted by establishing calibration model. The correlationcoefficients of hematocrit and hemoglobin can reach0.81and0.73relatively, with theRMSEP being1.82%and14.53g·L-1.4) A noise reduction processing method for pulse wave signal based on adaptivefilter was proposed by researching reduction processing of pulse wave signal noisedeeply. It was shown that the correlation coefficients of hematocrit and hemoglobinwere up to0.87and0.83relatively, with the RMSEP being1.89%and9.16g·L-1byusing the pulse wave signal after processing.This paper focused on the key problems of monitoring fast and high SNR signalfor near-infrared non-invasive biochemical analysis. And a fast and high SNRacquisition system of near-infrared volume pulse wave was explored. The researchesof this paper provide both theoretical and experimental basis of the clinicalapplication of NIRS non-invasive biochemical monitoring method. |
PDF Full Text Request