Comparing baseline correction algorithms in discriminating brownish soils from five proximity locations based on UPLC...

Area of Science:

• Forensic Science
• Analytical Chemistry
• Chemometrics

Background:

• Soil analysis is crucial for forensic investigations, linking suspects and victims to crime scenes.
• Ultra-Performance Liquid Chromatography (UPLC) is used for soil chemical profiling.
• UPLC chromatograms often suffer from unstable baselines, hindering accurate analysis.

Purpose of the Study:

• To compare the effectiveness of five baseline correction (BC) algorithms for UPLC soil data.
• To identify the optimal BC method for improving the discrimination of soil samples from different locations.
• To enhance the reliability of UPLC-based forensic soil analysis.

Main Methods:

• Five BC algorithms were evaluated: asymmetric least squares (AsLS), fill peak, iterative restricted least squares, median window (MW), and modified polynomial fitting.
• Thirty UPLC chromatograms from five distinct soil locations were analyzed.
• Principal Component Analysis (PCA) and Partial Least Squares-Discriminant Analysis (PLS-DA) were used to assess data quality and model performance.

Main Results:

• Raw UPLC soil chromatograms exhibited significant baseline fluctuations.
• Asymmetric least squares (AsLS) and median window (MW) showed the most substantial baseline correction.
• Principal Component Analysis (PCA) scores plots indicated improved data clustering after BC.
• Partial Least Squares-Discriminant Analysis (PLS-DA) models demonstrated that MW achieved the highest mean prediction accuracy.

Conclusions:

• Median window (MW) is the most effective baseline correction algorithm for UPLC soil data.
• MW significantly improves the quality and discriminatory power of UPLC chromatograms for forensic soil analysis.
• This study provides a robust method for enhancing forensic soil profiling using UPLC data.