Reduction of the influence of film thickness on diffuse reflectance spectroscopy measurement of the tongue.

Abstract

Diffuse reflectance spectroscopy of the tongue contains rich physiological and biochemical information of human status. As it involves contact measurement of the tongue, preventing cross infection during measurement is an important problem to be solved. The use of sterile isolation films is a convenient and efficient way. However, these films can affect the spectral measurement results at the same time. In this paper, to study the influence of films on spectral analysis, a new modeling strategy with good robustness is proposed. The model is established by the use of spectra obtained with different thicknesses of films; thus it is insensitive to the film thickness. The experiment was conducted by gathering spectra of 35 volunteers' tongue using films of different thicknesses. First, the partial least squares method was used to reveal the relationship between the spectral data and the thickness of films. Second, the support vector machine established a classification model between the spectral data and subjects to study the influence of the film. The results showed that the model established by the new modeling strategy could minimize the error caused by variation of the film thickness.