Noninvasive Testing Of Blood Alcohol Concentration Of Body Based On Near-infrared Spectroscopy And Its Key Technology

The near-infrared spectroscopy (NIRS) signals of alcohol in vivo are always contaminated by noise. In this study, wavelet analysis was used to eliminate noise and thereby detecting the NIRS signals of alcohol in vivo. Three different ways of wavelet transform de-noising were used. Signal noise ratio (SNR) and root mean square error (RMSE) method were used to evaluate the effects of the de-noising. The results show that the default threshold de-noising has the best effects. Therefore, the default threshold de-noising was chosen to perform de-noising analysis in vivo and in vitro. Our result shows that the wavelet transform de-noising is effective in removing noise from NRS signals of alcohol in vivo. With different alcohol concentration, the de-noising spectrums can show evident absorb peaks.To study the metabolism law of blood alcohol concentration(BAC) in different states. Through strictly control the experimental variables, let different types of experimental subjects (alcohol sensitive type, alcohol ordinary type) drink different dose alcohol (small-dose:0.4g/kg, normal-dose:0.6g/kg, over-dose:0.75g/kg, excess over-dose:0.9g/kg) at different state (emptiness, half-emptiness, full), the BAC was measured. The model of BAC over time was established, and the typical BAC change curve will be got.The main absorption peaks of alcohol were found at1670nm-1750nm and2230nm-2380nm from alcoholic solution spectrum. A quantitative model was established with a principal component using partial least squares (PLS) of the spectrum and BAC. The results show that it has good prediction accuracy. The R-square=0.956, σ=6.54,RMSE=42.7768of PLS regression model. Near-infrared spectroscopy might provide a new method to the measurement of alcohol in vivo.A three structures layers (epidermis, dermis, subcutaneous) optical model of human skin was built by Monte-Carlo simulation and measurement experiments at the wavelength range of1000～2000nm and a group of optimal optical parameter was chosen from previous research papers’data. The multi-layers optical model was used to study photons propagation laws in skin tissue using Monte-Carlo method. The photons propagation laws include the relationships among emitting photons density, photon propagation average depth, photon propagation length at each layer, wavelength and detection distance. The results showed that the relationships were strongly influenced by absorption coefficient. Photon propagation length in dermis and subcutaneous had a negative correlation with the absorption coefficient, as well as emitting photons density and photon propagation average depth.The study of Monte-Carlo simulation proves a theoretical basis for the optimal testing distance and noninvasive testing of human BAC. And some key technologies are designed, that include the integrated design of the optical spectrum instrument, light source, the design of detector and optical fiber probe for alcohol non-invasive testing.