End-to-end reconstruction of OCT optical properties and speckle-reduced structural intensity via physics-based learning

TL;DR

This work tackles the OCT inverse scattering problem of recovering intrinsic tissue optical properties ($n$, $\mu_s$, $g$) and speckle-reduced structural information from intensity data. It introduces a physics-informed end-to-end deep learning framework that embeds a differentiable Extended Huygens–Fresnel forward theory and diffusion-based score priors, trained on Monte Carlo corneal simulations. The approach jointly estimates optical-property maps and a cleaned structural intensity in a single pass, achieving robust performance and improved layer visualization under noise. By integrating physics-based constraints with learned priors, the method enables quantitative, label-free tissue characterization in OCT and demonstrates the value of physics-informed DL for computational OCT.

Abstract

Inverse scattering in optical coherence tomography (OCT) seeks to recover both structural images and intrinsic tissue optical properties, including refractive index, scattering coefficient, and anisotropy. This inverse problem is challenging due to attenuation, speckle noise, and strong coupling among parameters. We propose a regularized end-to-end deep learning framework that jointly reconstructs optical parameter maps and speckle-reduced OCT structural intensity for layer visualization. Trained with Monte Carlo-simulated ground truth, our network incorporates a physics-based OCT forward model that generates predicted signals from the estimated parameters, providing physics-consistent supervision for parameter recovery and artifact suppression. Experiments on the synthetic corneal OCT dataset demonstrate robust optical map recovery under noise, improved resolution, and enhanced structural fidelity. This approach enables quantitative multi-parameter tissue characterization and highlights the benefit of combining physics-informed modeling with deep learning for computational OCT.

Figures (2)

• Figure 1: End-to-end physics-regularized framework for OCT inverse scattering. Red box: U-Net predictors used for both training and testing; blue box: loss modules applied only during training.
• Figure 2: End-to-end reconstruction of OCT structural intensity and quantitative optical maps with ablation studies.