Sparse NIR optimization method (SNIRO) to quantify analyte composition with visible (VIS)/near infrared (NIR)...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Biorefinery

Background:

• High-throughput phenotyping relies on Visual-Near-Infra-Red (VIS/NIR) spectroscopy for analyzing organic materials.
• Current VIS/NIR spectrophotometers are often costly and too large for field use, limiting their practical application.

Purpose of the Study:

• To develop a novel method, Sparse NIR Optimization (SNIRO), for selecting optimal wavelengths for analyte quantification.
• To evaluate SNIRO's efficiency and accuracy against existing methods for analyzing various organic samples.

Main Methods:

• Developed the Sparse NIR Optimization (SNIRO) method using linear regression to select a subset of wavelengths.
• Compared SNIRO's computational time and accuracy with Marten's test, forward selection, and LASSO.
• Applied methods to publicly available datasets for protein content in corn flour and meat, and octane number in diesel.

Main Results:

• SNIRO demonstrated comparable accuracy to existing methods while potentially offering computational advantages.
• Successfully determined glucose content in Ulva sp. seaweed, a novel application for SNIRO.
• Validated SNIRO's effectiveness across diverse organic materials and analytes.

Conclusions:

• SNIRO provides an efficient approach for wavelength selection in VIS/NIR spectroscopy.
• The method facilitates the design of cost-effective, compact spectrophotometers for field-operable content analysis.
• SNIRO advances the potential for on-site analysis of complex organic materials, including novel feedstocks like Ulva sp.