Pseudo-time particle filtering for diffuse optical tomography.

Area of Science:

• Biomedical optics
• Medical imaging
• Computational physics

Background:

• Diffuse Optical Tomography (DOT) reconstructs optical properties of biological tissues.
• Traditional DOT methods often face challenges with ill-posedness and computational demands.
• Deterministic algorithms can struggle with accuracy in regions of low measurement sensitivity.

Purpose of the Study:

• To develop and evaluate a novel pseudo-dynamical framework for DOT reconstruction.
• To apply particle filters, specifically bootstrap (BS) and Gaussian-sum (GS) filters, for recovering optical parameters.
• To assess the accuracy and efficiency of these filters in reconstructing optical absorption coefficient distributions.

Main Methods:

• Recasting DOT reconstruction as a pseudo-dynamical problem.
• Implementing and utilizing BS and GS particle filters for parameter recovery.
• Employing indicator functions or continuous kernels to represent inclusions, reducing parameter space.
• Using numerically simulated and experimentally generated photon fluence data for validation.

Main Results:

• Both BS and GS filters provided accurate recovery of inclusion height, radius, and location.
• The GS filter demonstrated superior accuracy compared to the BS filter.
• Accurate contrast recovery was achieved even in the object's interior, overcoming limitations of deterministic methods.
• Reconstructed solutions were verified as admissible by comparing computed data with input data.

Conclusions:

• Particle filters offer a robust approach for diffuse optical tomography reconstruction.
• The pseudo-dynamical framework with particle filters effectively addresses the ill-posed nature of DOT.
• Reduced parameterization significantly improves computational efficiency for DOT.
• This method shows promise for accurate imaging in challenging biological tissues.