Concentration detection in turbid solutions based on volume scattering and dual-modality spectral fusion

View at publisher Summarise with VIRAT
Authors: Jianxin Zhang, Ming Wang, Xinru Zhang, Yun Xu
Cited by: 0
References: 32

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Abstract The scattering properties of turbid solutions are essential for concentration detection. To address the limitations of existing single-modality methods, which fail to effectively capture full-band scattering and extinction characteristics across a wide concentration range, this study proposes a concentration detection method based on volume scattering and dual-modality spectral fusion. A dual-modality spectral acquisition system was developed to simultaneously capture transmitted light spots and scattered beam images from a rectangular sample cell illuminated by a laser across multiple wavelength bands, thereby acquiring complementary optical information to enhance both the accuracy and range of turbid solution concentration detection. Experiments were conducted using Intralipid-20% samples at varying concentrations. Radial intensity features sensitive to extinction were extracted from the transmission side within the effective band, while beam morphology features in the 500–800 nm range were quantified using spatial moments on the scattering side. Partial least squares regression models were established based on transmission modality, scattering modality, and concatenated dual-modality features. Decision-level fusion was further implemented via multiple linear regression. The performance of the dual-modality fusion method was systematically compared with that of various single-modality approaches. Results show that the proposed method, by integrating complementary information from transmission intensity and scattered beam spatial morphology through a single synchronous imaging step, achieved a coefficient of determination ( R 2 ) of 0.9922, a root mean square error of 0.0148, and a ratio of performance to deviation of 11.6229 on the cross-validated test set. This approach significantly improves prediction accuracy across a broad concentration range by effectively fusing spatial morphological features with spectral intensity information.

Fields of Study & Subjects

Optical Imaging and Spectroscopy Techniques Spectroscopy and Chemometric Analyses Water Quality Monitoring and Analysis

Publication details

Journal / source
Measurement Science and Technology
Volume
37
Issue
14
Pages
145501–145501
Publication date
25 March 2026
Language
EN