Search

Highlights • The study designed and developed a heteromorphic sample pool. • Modeling method based on 2D light intensity distribution patterns can effectively enhance components detection precision and speed. • The study can provide meaningful and insight guidance for analytical chemistry and high-throughput test. Abstract To enhance the detection precision of samples with scattering characteristics by near infrared spectroscopy (NIRS), this study developed a heteromorphic sample pool and established the related 2D light intensity acquisition system, which can simultaneously acquire multi-path exit light adsorption and scattering information of the samples under test. The Intralipid-20% phantom solutions in 34 samples with different concentrations were detected, while one-dimensional (1D) exit light intensity distributions and two-dimensional (2D) exit light intensity distributions on the surface of the samples were analyzed and modeled using partial least squares. In contrast with the prediction results based on the modeling method of 1D exit light intensity distribution, the modeling method of 2D exit light intensity distribution exhibits more favorable results; specifically, correlation coefficient enhanced by 2.48%, while root mean square error reduced by 6.89%. The experimental results demonstrate that using heteromorphic sample pool can effectively achieve NIRS-based detection precision and speed of chemical components in the solutions with scattering characteristics, which can provide important references for high-throughput and high-precision detection of turbid media in analytical chemistry. [ABSTRACT FROM AUTHOR]

• In this study, a visualized light intensity distribution image is constructed by multi-dimensional spectroscopy fusion method, which extends the wavelength dimension of the model, reduces the data volume of the model, and further improves the prediction accuracy of the model. • The multi-dimensional spectroscopy fusion method of the wedge-shaped sample pool can accurately obtain the component content of the measured complex solution and realize the on-line, non-destructive and rapid detection of the complex solution. In order to improve the accuracy of the detection of complex solution components in wedge-shaped sample pool, this paper obtained multi-dimensional spectroscopy (MDS) of multi-path non-linear features including samples by multi-position and multi-wavelength detection methods. And this research constructed a visualized light intensity distribution image by multi-dimensional spectroscopy fusion (MDSF) method. The detection of 46 samples of India Ink and Intralipd-20% mixed phantoms was carried out, and the visualized light intensity distribution image was analyzed and trained by partial least squares (PLS) method, and compared to training method that the raw data without the MDSF (Non-MDSF). In this study, the multi-dimensional spectral fusion modeling method is used to increase the correlation coefficient of the model by 1.29%, and the mean square error is reduced by 3.07%. The results show that the use of MDSF method to analyze the composition of complex solutions, the composition of the solution with scattering characteristics is fast and accurate, which is of great significance for the field of complex solution detection and analysis. [ABSTRACT FROM AUTHOR]

• This paper used wedge sample as a detection model. • This paper studied the changes of the diffuse disc properties. • This paper proposed the multi-dimension radial distance method. The measurement accuracy of turbid media is difficult to improve due to the co-existence of absorption and scattering effects. In this report, the diffuse disc of transmission surface was obtained by using the Monte Carlo simulation model based on wedge sample, and the radial distance of each diffuse disc was extracted to study the influence of different optical parameters on the diffuse disc properties. Subsequently, the partial least square regression (PLS) analysis model was established to analyze the relationship between multi-dimensional radial distance and the concentration of Intralipid-20%. Compared with multi-dimensional spectroscopy fusion (MDSF) and conventional two-dimensional light intensity method, multidimensional radial distance (MDRD) method increased the average correlation coefficient of prediction set by 3.5% and decreased the average mean square error by 28.6%. This study proposes the high-throughput and non-invasive method to analysis the turbid media using Monte Carlo simulation model based wedge sample. [ABSTRACT FROM AUTHOR]