Hyperspectral image analysis for CARS, SRS, and Raman data

Area of Science:

• Spectroscopy
• Chemical Analysis
• Image Processing

Background:

• Hyperspectral image analysis (HIA) enables quantitative chemical composition analysis by factorizing data into component concentrations and spectra.
• Previous HIA methods had limitations in handling spatial variations and data acquisition speed.

Purpose of the Study:

• To significantly enhance the capabilities of hyperspectral image analysis (HIA).
• To improve the accuracy and efficiency of quantitative chemical composition analysis from hyperspectral data.

Main Methods:

• Implemented spatial weighting to minimize spectral error variations.
• Introduced a new spectral selection criterion for sparse sampling to accelerate hyperspectral imaging.
• Developed a singular value decomposition filter to remove data outliers and motion artifacts.

Main Results:

• Enhanced HIA shows improved retrieval of chemical components, especially those with local but small global concentrations.
• Sparse sampling with the new criterion significantly speeds up sequential hyperspectral imaging.
• Outlier filtering effectively suppresses artifacts like motion, improving data reliability.

Conclusions:

• The enhanced HIA offers improved accuracy and efficiency for quantitative chemical analysis.
• The developed methods are applicable to various Raman spectroscopy techniques.
• The HIA software is publicly available for use.