Comparison of three algorithms for the baseline correction of hyphenated data objects

Area of Science:

• Analytical Chemistry
• Chemometrics

Background:

• Baseline drift is a common artifact in chromatographic and mass spectrometry data.
• Accurate baseline correction is crucial for reliable data analysis and interpretation.
• Novel methods are needed to address the complexities of two-way data from hyphenated techniques.

Purpose of the Study:

• To evaluate three two-way baseline correction algorithms: orthogonal basis (OB), fuzzy optimal associative memory (FOAM), and polynomial fitting (PF).
• To assess the impact of these algorithms on signal-to-noise ratios (SNRs) in HPLC-MS data.
• To determine the effect of baseline correction on pattern recognition accuracy in GC/MS data.

Main Methods:

• Comparison of OB and FOAM (two-way) with PF (pseudo-two-way) algorithms.
• Evaluation using high-performance liquid chromatography-mass spectrometry (HPLC-MS) and gas chromatography/mass spectrometry (GC/MS) data.
• Assessment of SNRs on total ion current (TIC) chromatograms and classification accuracy using fuzzy rule-building expert system (FuRES) and partial least-squares-discriminant analysis (PLS-DA).

Main Results:

• Baseline correction, particularly with OB and FOAM, significantly increased SNRs of major peaks in HPLC-MS TIC chromatograms.
• Applying baseline correction to GC/MS data prior to pattern recognition enhanced prediction accuracies.
• Data transformation was found to further improve the performance of baseline correction methods.

Conclusions:

• The novel in-house algorithms, OB and FOAM, are effective for two-way baseline correction in hyphenated chromatography/mass spectrometry.
• Polynomial fitting (PF) is a viable, conventional method for one-way data, adapted for two-way data when blank objects are unavailable.
• Baseline correction is essential for improving data quality and enhancing chemometric analysis of complex chromatographic-mass spectrometry datasets.