Hyperspectral confocal fluorescence imaging: exploring alternative multivariate curve resolution approaches.

Area of Science:

• Spectroscopy
• Microscopy
• Data Analysis

Background:

• Hyperspectral confocal fluorescence microscopy enables quantitative imaging of multi-fluorophore samples.
• Multivariate curve resolution (MCR) is commonly used for analyzing hyperspectral images.
• Fully non-negatively constrained MCR models are standard due to expected non-negative spectral properties.

Purpose of the Study:

• To demonstrate the advantages of partially constrained MCR models over fully constrained models for hyperspectral image analysis.
• To show improved interpretability of MCR solutions in specific complex scenarios.

Main Methods:

• Application of constrained alternating least squares MCR analysis.
• Comparison of fully non-negatively constrained MCR models with partially constrained MCR models.
• Analysis of hyperspectral images from diverse samples: spherical beads, lung cells, quantum dots, and corn seed tissues.

Main Results:

• Partially constrained MCR models provided superior interpretability in four distinct cases.
• These models effectively handled system artifacts and minor spectral perturbations.
• Environmental effects and geometric changes were better described by derivative components in partially constrained solutions.

Conclusions:

• Partially constrained MCR models are crucial for accurate analysis of hyperspectral images with artifacts or environmental variations.
• The study highlights the need to explore alternative partially constrained models beyond standard fully constrained approaches.
• This work advances quantitative imaging techniques in fluorescence microscopy.