Gradient-based quantitative image reconstruction in ultrasound-modulated optical tomography: first harmonic measurement type in a linearised diffusion formulation

TL;DR

This paper introduces a gradient-based image reconstruction method for ultrasound-modulated optical tomography, utilizing a linearised diffusion model and adjoint techniques to efficiently recover optical properties with high accuracy.

Contribution

It develops a novel linearised diffusion model for first-harmonic measurements and employs an adjoint-assisted gradient method for efficient, accurate image reconstruction.

Findings

Successful recovery of optical absorption and scattering parameters within +/-5% accuracy.

Validation with simulated noisy data demonstrates robustness and effectiveness.

Reconstruction in both 2D and 3D confirms method's applicability.

Abstract

Ultrasound-modulated optical tomography is an emerging biomedical imaging modality which uses the spatially localised acoustically-driven modulation of coherent light as a probe of the structure and optical properties of biological tissues. In this work we begin by providing an overview of forward modelling methods, before deriving a linearised diffusion-style model which calculates the first-harmonic modulated flux measured on the boundary of a given domain. We derive and examine the correlation measurement density functions of the model which describe the sensitivity of the modality to perturbations in the optical parameters of interest. Finally, we employ said functions in the development of an adjoint-assisted gradient based image reconstruction method, which ameliorates the computational burden and memory requirements of a traditional Newton-based optimisation approach. We validate our work by performing reconstructions of optical absorption and scattering in two- and three-dimensions using simulated measurements with 1% proportional Gaussian noise, and demonstrate the successful recovery of the parameters to within +/-5% of their true values when the resolution of the ultrasound raster probing the domain is sufficient to delineate perturbing inclusions.