SSASU

SSASU performs semi-blind sparse affine spectral unmixing to separate fluorophore signals from autofluorescence and background in spectral micrographs, improving accuracy of fluorophore concentration estimation.

Key Features:

• Affine Mixture Model (AMM): Incorporates background fluorescence into the mixing model as a generalization of the linear mixing model.
• Affine Nonnegative Matrix Factorization: Estimates fluorophore endmember spectra from reference images using an affine nonnegative matrix factorization approach.
• Semi-blind Sparse Affine Spectral Unmixing Algorithm: Learns and adaptively separates autofluorescence and background spectra on a per-image basis to determine fluorophore concentrations.
• Reduced Proportion Indeterminacy: Lowers proportion indeterminacy relative to existing methods while maintaining a given relative reconstruction error in autofluorescence-contaminated spectral images.

Scientific Applications:

• Cellular and Molecular Biology: Imaging of cellular components tagged with multiple fluorophores.
• Pathology and Medical Diagnostics: Detection and quantification of biomarkers in autofluorescent tissue samples.
• Neuroscience: Analysis of complex fluorescent labeling to map neural circuits.

Methodology:

Adopts the Affine Mixture Model to represent background fluorescence; applies affine nonnegative matrix factorization to estimate fluorophore endmember spectra from reference images; and executes a semi-blind sparse affine spectral unmixing algorithm that adaptively learns and separates autofluorescence and background spectra on a per-image basis.