Baseline correction combined partial least squares algorithm and its application in on-line Fourier transform...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Fourier Transform Infrared (FTIR) spectroscopy is a powerful analytical technique.
• Baseline drift and polynomial interferences can significantly affect quantitative analysis in FTIR, particularly in Attenuated Total Reflectance (ATR-FTIR).
• Conventional Partial Least Squares (PLS) regression can be sensitive to spectral baseline variations.

Purpose of the Study:

• To develop a novel quantitative calibration algorithm for ATR-FTIR spectroscopy that effectively addresses baseline interferences.
• To improve the accuracy and reliability of online quantitative analysis using ATR-FTIR data.
• To introduce a method that integrates baseline correction directly into the PLS model building process.

Main Methods:

• A new algorithm, Baseline Correction Combined Partial Least Squares (BCC-PLS), was developed.
• The BCC-PLS algorithm embeds baseline correction constraints into the PLS weights selection process.
• The algorithm's performance was validated using Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR) spectra of glucose and marzipan samples.

Main Results:

• The BCC-PLS algorithm demonstrated superior prediction performance compared to conventional PLS.
• Accurate prediction of moisture content in marzipan was achieved with a Root Mean Square Error of Cross-Validation (RMSECV) of 0.53% (w/w).
• Sugar content in marzipan was predicted with an RMSECV of 2.04% (w/w).

Conclusions:

• The proposed BCC-PLS algorithm effectively eliminates lower-order polynomial interferences and overcomes baseline uncertainty in ATR-FTIR quantitative analysis.
• BCC-PLS is suitable for on-line quantitative analysis requirements, offering improved accuracy.
• The algorithm provides a robust solution for the quantitative determination of components in complex samples like marzipan using ATR-FTIR.