Calibration-free 3D imaging through scattering media via diffusion-kernel-guided reconstruction

Area of Science:

• Biomedical Optics
• Image Reconstruction
• Diffusion Imaging

Background:

• Three-dimensional (3D) imaging using the diffusion equation often requires knowing the medium's thickness and optical properties.
• This dependency limits the practical application of existing 3D imaging techniques.

Purpose of the Study:

• To develop a calibration-free 3D imaging framework that bypasses the need for prior knowledge of medium properties.
• To enable robust reconstruction of hidden targets in scattering media.

Main Methods:

• A semi-empirical forward model was formulated to capture both backscattered and target signals.
• A multi-stage optimization strategy was used to estimate medium parameters and decouple the target signal.
• Diffusion-kernel-guided spatially adaptive regularization was employed for image reconstruction.

Main Results:

• The proposed method successfully estimates medium parameters and decouples the target signal without calibration.
• Spatially adaptive regularization effectively suppresses noise while preserving target structural features.
• Experiments demonstrated robust reconstruction performance across various medium thicknesses and multi-target scenarios.

Conclusions:

• The developed framework offers a practical solution for 3D imaging in scattering media.
• This calibration-free approach significantly enhances the applicability of diffusion-based imaging.
• The method shows promise for applications requiring non-invasive imaging through turbid environments.