Time-domain feature extraction for target specificity in photoacoustic remote sensing microscopy

Area of Science:

• Biomedical Optics
• Microscopy Techniques
• Biophotonics

Background:

• Photoacoustic remote sensing (PARS) microscopy is a label-free imaging technique.
• PARS captures optical fluctuations from photoacoustic pressures.
• Current methods primarily use signal amplitude, overlooking material property information within time-domain (TD) signals.

Purpose of the Study:

• To develop a novel method for extracting material property information from PARS TD signals.
• To utilize signal shape-based clustering for enhanced tissue characterization.
• To enable label-free differentiation of tissue constituents using PARS.

Main Methods:

• A novel clustering method, a modified K-means algorithm, was applied to TD signals.
• The clustering focused on learning features based on signal shape.
• These learned features were used to create virtual colorizations of tissue.

Main Results:

• The clustering method successfully identified and clustered TD signals based on shape.
• Virtual colorizations highlighted distinct regions within fresh murine brain tissue.
• These visualizations correlated with underlying material properties.

Conclusions:

• Signal shape analysis in PARS TD data can reveal underlying material properties.
• This approach offers a new pathway for label-free tissue differentiation.
• Potential applications include distinguishing myelinated/unmyelinated axons and cell nuclei.