Unmixing biological fluorescence image data with sparse and low-rank Poisson regression

Area of Science:

• Biological imaging
• Microscopy
• Spectroscopy

Background:

• Multispectral fluorescence microscopy identifies multiple targets in complex biological samples.
• Accuracy of fluorophore unmixing decreases with increased fluorophore numbers and lower signal-to-noise ratios.
• Spatial distribution information can improve fluorophore identification accuracy.

Purpose of the Study:

• To develop an improved method for deconvolution of spectral images with highly overlapping fluorophores.
• To address challenges in low signal-to-noise regimes within biological fluorescence microscopy.
• To enhance the accuracy of fluorophore identification and abundance estimation.

Main Methods:

• Proposed a regularized sparse and low-rank Poisson regression unmixing approach (SL-PRU).
• Implemented multipenalty terms for sparseness and spatial correlation.
• Utilized Poisson regression for photon abundance estimation and developed a parameter tuning method.

Main Results:

• SL-PRU demonstrated improved accuracy in unmixing fluorophores with highly overlapping spectra.
• The method effectively handles low signal-to-noise ratios in recorded images.
• Validation on simulated and real-world images confirmed the method's efficacy.

Conclusions:

• The SL-PRU method offers enhanced accuracy for multispectral fluorescence microscopy.
• This approach is valuable for analyzing complex biological samples with spectral overlaps.
• The developed algorithm improves the reliability of fluorophore identification in challenging imaging conditions.