Autofluorescence removal by non-negative matrix factorization.

Area of Science:

• Microscopy
• Biophotonics
• Image Analysis

Background:

• Autofluorescence (AF) in biological tissues confounds accurate measurement of fluorescent reporter molecules in microscopy.
• Existing spectral unmixing methods for AF removal often require prior knowledge or yield physically implausible negative coefficients.

Purpose of the Study:

• To develop a novel, physically interpretable, and fully automatic algorithm for tissue autofluorescence removal.
• To introduce a new application of non-negative matrix factorization (NMF) for AF removal in fluorescence microscopy.

Main Methods:

• A new non-negative matrix factorization (NMF) algorithm was developed for spectral unmixing.
• The algorithm separates true fluorescence signals from autofluorescence components by estimating dark current.
• A test-bed using fluorescent beads was created to evaluate AF removal algorithm performance.

Main Results:

• The novel NMF algorithm successfully separated true signal and autofluorescence components.
• The developed algorithm demonstrated superior performance compared to existing state-of-the-art methods on validation images.
• The algorithm provides physically meaningful, non-negative mixing coefficients.

Conclusions:

• The new NMF-based algorithm offers an effective and automatic solution for tissue autofluorescence removal in fluorescence microscopy.
• This method enhances the accuracy of quantitative analysis of fluorescent reporter molecules in biological samples.
• The algorithm represents a significant advancement over previous AF removal techniques.