Critical comparison of background correction algorithms used in chromatography

Area of Science:

• Analytical Chemistry
• Chromatography Data Analysis

Background:

• Drift and noise significantly impact chromatogram analysis accuracy.
• Existing drift and noise-removal algorithms lack standardized, comparative performance data.

Purpose of the Study:

• To quantitatively compare various drift and noise-removal algorithms on a standardized basis.
• To develop a robust data generation tool for rigorous algorithm assessment.
• To identify optimal algorithm combinations for different signal-to-noise ratios.

Main Methods:

• Developed a data generation tool using experimental backgrounds and peak shapes.
• Created hybrid datasets with known peak profiles, areas, and backgrounds (500 chromatograms).
• Evaluated 35 combinations of seven drift-correction and five noise-removal algorithms.

Main Results:

• Sparsity-assisted signal smoothing with asymmetrically reweighted penalized least-squares minimized errors in low-noise signals.
• For high-noise signals, sparsity-assisted signal smoothing and local minimum background correction yielded lower absolute peak area errors.
• Algorithm performance varied with peak density, background shape, and noise levels.

Conclusions:

• The developed data generation tool enables fair and rigorous comparison of signal processing algorithms.
• Specific algorithm combinations are recommended based on signal noise characteristics.
• Findings support the automation of chromatographic data analysis workflows.