Enhancing biomechanical stimulated Brillouin scattering imaging with physics-driven model selection

TL;DR

This paper introduces a physics-driven model selection framework for Brillouin microscopy, improving spectral artifact reduction and enabling better differentiation of single- and multi-peak signatures in biomechanical imaging.

Contribution

The study presents a novel model selection approach based on information theory and water peak thresholds to enhance Brillouin shift quantification in microscopy.

Findings

Reduced spectral artifacts in Brillouin images

Improved differentiation of single- and multi-peak signatures

Enhanced quantification of biomechanical properties

Abstract

Brillouin microscopy is an emerging technique for all-optical biomechanical imaging without the need for physical contact with the sample or for an external mechanical stimulus. However, Brillouin microscopy often retrieves a single, averaged Brillouin frequency shift of all the materials in the sampling volume, introducing significant spectral artifacts in the Brillouin shift images produced. To enable the identification between single- and multi-peak Brillouin signatures in the sample voxels, we developed here a new physics-driven model selection framework based on information theory and an overfit Brillouin water peak threshold. The model selection framework was applied to Brillouin data of NIH/3T3 cells measured by stimulated Brillouin scattering microscopy, facilitating the improved quantification of the Brillouin shift of different regions in the cells, and substantially minimizing spectral artifacts in their Brillouin shift images.