Methods To Improve The Analysis Accuracy Of Spectroscopy And Its Application In Tongue Diagnose

Spectroscopy could reflect composition,structure and status information of the measured substance,therefore,it has caused widespread attention in medical healthcare and other fields.However,the spectral measurement essentially belongs to weak signal detection,which is easily affected by many factors,such as individual differences,instrument noise,measurement environments and so on.Therefore,it's the key to suppress the influence of these factors to improve the measurement and analysis accuracy.The following work has been done in order to reduce the inference of ambient light,dark current,and the inconsistency of the film in the tongue diagnose.(1)The effect of ambient light and dark current should not be neglected in spectroscopic measurement.The optical chopper is applied in spectral data collection system to suppress the influence.Based on the modulate chopper technology,principal frequency component analysis(PFCA)method is proposed to further improve the measurement accuracy.The feasibility of the PFCA method is proved in theory and the corresponding solution experiments are designed to verify the effectiveness of the PFCA method.The results show that the PFCA method not only satisfactorily eliminates the interference signals but also extracts useful information as much as possible,improving the analysis accuracy.(2)In the modulate chopper technology,the spectrometer sampling rate is out of sync with the optical chopper frequency.Therefore,this paper studies the relationship between the two.MATLAB is used to simulate the process of chopper modulation and demodulation,and the suitable functional relationship has been acquired.The research widens the application of the chopper modulate technology and provides an important reference for the following configuration parameters.(3)In order to reduce the error of the inconsistency of the film in contact measurement of the spectral tongue diagnosis,this paper takes the coverage modeling strategy of “M+N” theory.The modeling strategy makes no distinction between the N factor the thicknesses of the film and the M element the target components.A hybrid calibration set that both of two are uniformly distributed in the calibration set is built.The result shows that this modeling strategy could minimize the error caused by variation of film thickness.It addresses some obstacles for the following spectral tongue diagnosis in clinical trials.In this paper,to reduce or even eliminate the interference factors in the process of spectral tongue measurement,this paper put forward the corresponding methods.The feasibility and validity of these methods are verified in the theory and experiments.It also provides the reference for other related spectral signal detection.