Elastic net-based non-negative iterative three-operator splitting strategy for Cerenkov luminescence tomography

Area of Science:

• Biomedical optics
• Molecular imaging
• Medical physics

Background:

• Cerenkov luminescence tomography (CLT) is a key optical molecular imaging technique.
• Severe photon scattering in CLT leads to ill-posed inverse problems.
• Current CLT reconstruction accuracy and shape recovery are insufficient for clinical use.

Purpose of the Study:

• To enhance spatial location accuracy and shape recovery in CLT reconstructions.
• To address limitations of existing CLT imaging methods.

Main Methods:

• Proposed a non-negative iterative three operator splitting (NNITOS) strategy.
• Incorporated elastic net (EN) regularization for a non-convex optimization problem.
• Utilized L_1/2-norm regularization, adaptive grouping manifold learning, and region of interest shrinking.

Main Results:

• NNITOS demonstrated progressive convergence through iterative operator splitting.
• The method achieved superior location accuracy, shape recovery, and robustness compared to other techniques.
• Numerical simulations and in vivo experiments validated the effectiveness of NNITOS.

Conclusions:

• The NNITOS strategy significantly improves CLT reconstruction quality.
• This advancement holds promise for accelerating the clinical application of Cerenkov luminescence tomography.