Primal-dual approach to optical tomography with discretized path integral with efficient formulations

Area of Science:

• Medical Imaging
• Computational Science
• Applied Mathematics

Background:

• Optical tomography is a vital imaging technique.
• Reconstructing images in optical tomography involves solving complex inverse problems.
• Existing methods can be computationally intensive.

Purpose of the Study:

• To develop a more efficient optical tomography method.
• To improve the speed of image reconstruction in optical tomography.
• To maintain the accuracy of reconstructed images.

Main Methods:

• Introduced a primal-dual approach for the inverse problem.
• Formulated the problem as a constrained optimization task.
• Developed efficient computations for Jacobian and Hessian matrices.

Main Results:

• The proposed method is significantly faster ([Formula: see text]) than the log-barrier interior point method ([Formula: see text]).
• Achieved comparable image quality with a root-mean-square error increase of up to 12%.
• Demonstrated efficiency on the Shepp-Logan phantom with a grid size of [Formula: see text].

Conclusions:

• The discretized path integral approach offers a faster alternative for optical tomography.
• Efficient computation of optimization problem components is key to performance.
• This method balances speed and accuracy in optical image reconstruction.