Pre-processing and source separation methods for Raman spectra analysis of biomedical samples.

Area of Science:

• Biomedical Optics
• Spectroscopy
• Data Analysis

Background:

• Raman spectroscopy analyzes molecular composition in biological samples.
• Source separation methods can process complex Raman spectral data.
• Distortions like fluorescence and peak variations hinder source separation.

Purpose of the Study:

• To investigate the impact of preprocessing on source separation efficiency in Raman spectroscopy.
• To demonstrate improvements in analyzing biological samples, specifically human skin biopsies.

Main Methods:

• Applied preprocessing techniques to compensate for fluorescence, peak misalignment, and width heterogeneity.
• Utilized Independent Component Analysis (ICA) algorithm JADE.
• Employed positive source separation methods: Non-negative Matrix Factorization (NMF) and Maximum Likelihood Positive Source Separation (MLPSS).

Main Results:

• Preprocessing significantly enhances the efficiency of source separation methods for Raman spectra.
• Demonstrated successful numerical dewaxing of Raman signals from human skin biopsy.
• Showcased the dependency of source separation performance on the chosen preprocessing steps.

Conclusions:

• Effective preprocessing is essential for accurate molecular analysis using Raman spectroscopy.
• Source separation methods, when combined with appropriate preprocessing, offer powerful tools for biological sample analysis.
• The study highlights the critical role of data correction in advancing Raman spectral data interpretation.