L₁-L₂ Minimization Via A Proximal Operator For Fluorescence Molecular Tomography

Area of Science:

• Biomedical Imaging
• Medical Physics
• Computational Biology

Background:

• Fluorescent Molecular Tomography (FMT) offers sensitive, noninvasive 3D biomarker distribution imaging.
• Challenges in FMT include light scattering and ill-posed inverse problems, hindering accurate location and morphology reconstruction.
• Existing reconstruction methods face limitations in precision.

Purpose of the Study:

• To develop an improved reconstruction method for Fluorescent Molecular Tomography (FMT).
• To address the challenges of light scattering and inverse problem ill-posedness in FMT.
• To enhance the accuracy of biomarker localization and morphology in FMT.

Main Methods:

• Proposed L1-L2 norm regularization for FMT reconstruction.
• Utilized proximal operators of the non-convex L1-L2 norm.
• Employed the forward-backward splitting (FBS) method to solve the FMT inverse problem.

Main Results:

• The proposed FBS method demonstrated superior performance in simulation studies.
• The FBS method showed enhanced location accuracy compared to IVTCG, DCA, and IRW-L1/2 methods.
• The study confirmed the effectiveness of L1-L2 norm regularization in improving FMT reconstruction.

Conclusions:

• The L1-L2 norm regularization combined with the FBS method significantly improves FMT reconstruction accuracy.
• This approach offers a more precise tool for visualizing biomarker distribution in biological research.
• The developed method holds promise for advancing noninvasive molecular imaging techniques.