Nonconvex regularizations in fluorescence molecular tomography for sparsity enhancement

Area of Science:

• Preclinical Imaging
• Biomedical Engineering
• Medical Physics

Background:

• In vivo fluorescence imaging is a key functional imaging technique in preclinical research.
• Fluorescence molecular tomography (FMT) utilizes near-infrared probes for targeted tissue localization.
• Sparse solutions are preferred for accurate FMT image reconstruction.

Purpose of the Study:

• To investigate the efficacy of nonconvex regularization methods for enhancing sparsity in FMT.
• To compare the performance of novel nonconvex regularizations against standard L(1) regularization in small animal imaging.

Main Methods:

• Numerical simulations and phantom experiments were conducted for FMT image reconstruction.
• A majorization-minimization algorithm was employed for the iterative reconstruction process.
• Reconstructed images from nonconvex methods were compared with those from L(1) regularization.

Main Results:

• Proposed nonconvex regularization methods demonstrated superior performance in FMT.
• Accurate recovery of sparse targets was significantly enhanced using nonconvex approaches.
• Nonconvex methods outperformed L(1) regularization in precision and target localization.

Conclusions:

• Nonconvex regularization techniques offer significant advantages for FMT image reconstruction.
• These methods provide enhanced accuracy for sparse target recovery in small animal imaging.
• The findings support the adoption of nonconvex regularizations for advanced preclinical fluorescence imaging.