Preprocessing methods of Raman spectra for source extraction on biomedical samples: application on paraffin-embedded...

Area of Science:

• Spectroscopy
• Biomedical Optics
• Chemometrics

Background:

• Raman spectra are typically modeled as linear mixtures of constituent spectra.
• Instrumentation and biological sample properties introduce nonlinear distortions.
• These nonlinearities degrade the performance of source separation methods.

Purpose of the Study:

• To develop preprocessing techniques to correct nonlinear distortions in Raman spectra.
• To restore a linear model suitable for source separation analysis.
• To evaluate the impact of preprocessing on Raman spectral analysis.

Main Methods:

• Applied specific preprocessing steps to correct for dark current, detector/optic responses, fluorescence, and peak variations.
• Utilized two source separation methods: Joint Approximate Diagonalization of Eigenmatrices (JADE) and Maximum Likelihood Positive Source Separation (MLPSS).
• Tested methods on Raman spectral data from paraffin-embedded human skin biopsies.

Main Results:

• Preprocessing steps successfully corrected nonlinear distortions, enabling a linear spectral model.
• Source separation performance was significantly improved after applying the proposed preprocessing.
• The study demonstrated the efficacy of JADE and MLPSS on preprocessed biological Raman data.

Conclusions:

• Developed and validated preprocessing steps are crucial for accurate Raman spectral analysis of biological samples.
• Correcting nonlinearities enhances the reliability of source separation for identifying molecular constituents.
• The proposed methods offer a pathway to improved quantitative analysis using Raman spectroscopy in biomedical applications.