An efficient numerical method for general L(p) regularization in fluorescence molecular tomography.

Area of Science:

• Biomedical imaging
• Computational science
• Optical physics

Background:

• Fluorescence tomography faces challenges with ill-posed problems and large datasets.
• Existing reconstruction algorithms struggle with computational demands for 3D imaging.

Purpose of the Study:

• To develop an efficient reconstruction algorithm for fluorescence tomography.
• To incorporate general Lp regularization for improved accuracy and scalability.
• To address the challenges of ill-posedness and large-scale numerical problems in 3D imaging.

Main Methods:

• Designed and implemented a reconstruction algorithm using general Lp regularization (p ≥ 1).
• Employed an efficient matrix-free strategy to reduce memory and computational costs.
• Applied sparsity-promoting L1 constraints for experimental validation.

Main Results:

• The algorithm effectively handles large-scale fluorescence tomography problems.
• Matrix-free strategy significantly reduces computational and memory requirements.
• L1 regularization proved adequate for sparse phenomena in phantom experiments.

Conclusions:

• General Lp regularization, particularly L1, is a powerful tool for fluorescence tomography.
• The developed matrix-free algorithm offers an efficient solution for 3D imaging.
• This approach enhances the investigation of sparse phenomena in biological imaging.