Self-guided reconstruction for time-domain fluorescence molecular lifetime tomography

Area of Science:

• Biomedical Optics
• Image Reconstruction
• Fluorescence Imaging

Background:

• Fluorescence probes exhibit environment-dependent yields and lifetimes, complicating fluorescence molecular lifetime tomography (FMLT) reconstruction.
• Accurate reconstruction of heterogeneous targets in time-domain FMLT (TD-FMLT) remains a significant challenge.

Purpose of the Study:

• To introduce a novel algorithm, self-guided L1 regularization projected steepest descent (SGL1PSD), to enhance TD-FMLT reconstruction performance.
• To improve the accuracy of reconstructing fluorescence yield and lifetime maps for heterogeneous biological targets.

Main Methods:

• Developed the SGL1PSD algorithm, incorporating a time-resolved strategy for fluorescence yield reconstruction.
• Employed a four-step iterative approach: initial yield map, inverse lifetime map, refined yield map, and final inverse lifetime map.
• Utilized projected iterated Tikhonov regularization for yield maps and L1 regularization projected steepest descent for inverse lifetime maps.

Main Results:

• Phantom experiments with heterogeneous targets demonstrated the effectiveness of the SGL1PSD algorithm.
• The iterative refinement and time-resolved strategy significantly improved reconstruction accuracy compared to existing methods.
• The algorithm successfully reduced ill-posedness in high-dimensional nonlinear reconstruction problems.

Conclusions:

• The SGL1PSD algorithm offers a robust and accurate solution for TD-FMLT reconstruction of heterogeneous targets.
• The proposed time-resolved strategy is crucial for enhancing the fidelity of fluorescence yield and lifetime mapping.
• This advancement has the potential to improve diagnostic capabilities in fluorescence-guided imaging and molecular sensing.