Independent component analysis as a pretreatment method for parallel factor analysis to eliminate artefacts from...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Parallel factor analysis (PARAFAC) is widely applied to excitation-emission fluorescence data.
• Measurement artefacts like Rayleigh and Raman scattering can hinder PARAFAC component extraction and interpretation.
• Replacing spectral zones with missing values is not always optimal when informative signals overlap with artefacts.

Purpose of the Study:

• To introduce a novel method for removing scattering artefacts from fluorescence data.
• To improve the accuracy and interpretability of PARAFAC models.
• To demonstrate the utility of independent component analysis (ICA) in artefact removal.

Main Methods:

• Utilized independent component analysis (ICA) on the unfolded cubic array of fluorescence data.
• Identified and removed independent components associated with Rayleigh and Raman scattering.
• Reconstructed the excitation-emission fluorescence data cube after selective artefact removal.
• Applied PARAFAC to the artefact-corrected data.

Main Results:

• Successful identification and removal of scattering artefacts using ICA.
• Obtained satisfactory PARAFAC models after artefact removal.
• Demonstrated that the proposed method enhances the quality of fluorescence data analysis.
• Showcased the applicability of the method beyond 3D fluorescence data.

Conclusions:

• ICA is a powerful tool for removing scattering artefacts in fluorescence spectroscopy.
• The proposed selective artefact removal procedure improves PARAFAC analysis outcomes.
• This approach offers a robust alternative to traditional methods for handling spectral interferences.