Improving the nondestructive analysis accuracy of liquids in a flexible container based on the multi-pathlength spectrum method

In order to achieve nondestructive analysis of liquid samples, the spectral analysis of solution in a flexible container is proposed in this paper. However, the difference in the flexible containers reduces the accuracy of the spectral analysis. By analyzing the relationship between the condition number of coefficient matrix (CNCM) and the solution accuracy, a method of reducing CNCM by multi-pathlength spectrum is proposed to reduce the error caused by the differences in flexible containers. To verify the feasibility of the method, different concentrations of samples were prepared with intralipids and India-ink, and polyvinyl chloride sample tubes were used as the flexible container. The experimental results show that the proposed strategy can effectively reduce the error caused by the differences in flexible containers.

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

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.