Iterative Correction Scheme Based on Discrete Cosine Transform and L1 Regularization for Fluorescence Molecular...

Area of Science:

• Biomedical Imaging
• Optical Imaging
• Molecular Imaging

Background:

• High-intensity background fluorescence is a common challenge in fluorescence molecular tomography (FMT).
• Background fluorescence arises from fluorescent probe accumulation in non-target tissues or biological autofluorescence.
• This interference distorts reconstruction results, particularly with sparse fluorescent targets.

Purpose of the Study:

• To mitigate the detrimental effects of background fluorescence on FMT reconstruction.
• To enhance the accuracy and quality of FMT imaging in the presence of background noise.

Main Methods:

• An iterative correction algorithm is employed for FMT reconstruction.
• 3-D discrete cosine transform is used for filtering intermediate results within each iteration.
• L1 regularization is applied as a sparsity constraint to minimize objective function energy.

Main Results:

• Phantom experiments demonstrated improved reconstruction quality with the proposed iterative correction scheme.
• The method effectively reduced background fluorescence across various intensity levels and distributions.
• Validation was performed using homogeneous and heterogeneous background fluorescence conditions.

Conclusions:

• The developed iterative correction scheme effectively reduces background fluorescence in FMT.
• Filtering intermediate results and L1 regularization contribute to detail preservation and noise suppression.
• This approach significantly enhances the reliability and accuracy of FMT imaging.