Research On Several Critical Techniques Of The Non-invasive Measurement Of Blood Components Based On Dynamic Spectrum

Noninvasive blood components measurement can improve the modern laboratorymedicine technology and enhance the quality of people life. The near infraredtransmission spectroscopy as one of the methods, which were most likely to achievenon-invasive blood components measurement, has been got highly attention by theinternational researchers. The Dynamic Spectrum (DS) based on the near infraredPhotoelectric Plethysmography (PPG), can extract the blood absorption informationfrom the artery’s pulsation part and can eliminate the influence of the individualdifferences and measurement environment on the noninvasive blood componentsmeasurements effectively. This thesis is focused on studying the several criticaltechniques of achieving the noninvasive blood components measurement based on DStheory to improve the detection accuracy and promote the clinical application of thistechnology.The main contents of this thesis are follows:(1) Verified the feasibility of the DS technology for achieving the noninvasiveblood components measurement from the preliminary theoretical and experimentalapproach. Systematically and quantitatively analyzed and compared the SNR of DStime domain and frequency domain extraction methods. Presented a “performancetriangle” of the DS extraction methods, according to the accuracy of the DS extractionresults under the various interference for evaluating their overall performance.(2) Proposed the DS data quality evaluation criteria--effective wavelengths toguide the reasonable selection the sample data based on timely and effectively dataquality evaluation. The reasonable data selection procedure could improve thestability and accuracy of the prediction models. The experimental results showed thatthe average relative error of the hemoglobin, glucose and total cholesterol predictionvalue could decrease from the13.8%、15.8%、5.4%to6.5%、6.5%、2.1%after databeing evaluated and selected by the quality criteria, seperatively.(3) The Boxcar filter, the wavelet transform filtering and empirical modedecomposition filtering method were used to process the spectrum data andlogarithmic photoelectric pulse wave to eliminate interference noises and improve thequality of DS data. The light sources compensation and wavelength mergence approaches were used to enhance the SNR of spectrum, which could expand theeffective wavelength region of DS measurement and enhance the SNR of DS data.(4). A portable DS clinical acquisition system was established based on theVisual Basic program language. The system can achieve the spectrum acquisition, DSdata extraction, data processing and blood component prediction, etc. automatically,which increased the stability and efficiency of the DS data collection procedure. Anumber of clinical data acquisitions were conducted in the hospitals. The calibrationmodels between the DS data and hemoglobin, blood glucose, total cholesterolconcentration for quantitative analysis were established. The prediction results weresatisfied and close to the clinical request.The study demonstrated the feasibility of DS applied to noninvasive bloodcomponents measurement theoretically and experimentally. The SNR of DS data andthe accuracy of blood components prediction were improved based on the severalcritical techniques’ improvment and optimization. The experimental results weresatisfied, which showed the promising of DS method for achieving the bloodcomponents non-invasive measurement. This study has some significance on the DStheory development and further researches for achieving more blood componentsnoninvasive measurement.