Temporal Unmixing of Dynamic Fluorescent Images by Blind Source Separation Method with a Convex Framework

Area of Science:

• Biomedical Imaging
• Molecular Imaging
• Pharmacokinetics

Background:

• Dynamic fluorescence molecular tomography (FMT) enables in vivo studies of perfusion, biodistribution, and pharmacokinetics.
• Current methods reconstruct images frame-by-frame, followed by unmixing using PCA or ICA, which struggle with correlated sources.

Purpose of the Study:

• To develop and evaluate a temporal unmixing approach for dynamic FMT data.
• To overcome limitations of PCA and ICA in analyzing correlated kinetic patterns.

Main Methods:

• A temporal unmixing method based on nonnegative blind source separation (BSS) within a convex analysis framework was developed.
• The method deduces the relationship between measured imaging data and kinetic patterns.
• Numerical simulations and phantom experiments were used for validation.

Main Results:

• The proposed BSS-based method allows for feasible unmixing of measured data prior to tomographic reconstruction.
• It demonstrates superior unmixing quality compared to principal component analysis (PCA) and independent component analysis (ICA).
• The method does not assume source independence or zero correlations.

Conclusions:

• Temporal unmixing before tomographic reconstruction is a viable strategy for dynamic FMT.
• Nonnegative blind source separation offers improved performance for analyzing complex kinetic patterns in molecular imaging.