Algorithm for Improved Spectral Unmixing with Reference Image Integration

[sreekarreddydfci]

Feature Description:

A common issue in multiplex immunofluorescent imaging is the partial overlapping of fluorophore emission spectra in the captured image. Spectral unmixing is crucial in immunofluorescent imaging to handle overlapping emission spectra, address challenges posed by auto-fluorescence emission, and discern meaningful information from the individual fluorophores. The existing LUMoS implementation, which leverages k-means clustering, has provided a foundation for this unmixing process. However, in side-by-side comparisons, it falls short of the performance benchmark set by proprietary tools such as inForm. The goal of this feature request is to design an algorithm that makes use of reference images for each fluorophore, potentially increasing the reliability of unmixing.

Expected Outcome:

The implementation of such an algorithm is expected to elevate the performance of LUMoS, offering a reliable open-source alternative to commercial spectral unmixing tools.

Resources:

Current plugin implementation: LUMoS Bleedthrough Correction Plugin.