Chemical complexity of the retina addressed by novel phasor analysis of unstained multimodal microscopy

Abstract Unstained multimodal microscopy is capable of non-invasively obtaining chemically specific information with sub-micron spatial resolution; however, spectral overlap makes quantitative analysis difficult. Here, multimodal images that include second-harmonic generation, third-harmonic generation, and two- and three-photon excited fluorescence signals from unstained retinas are analyzed. The composition from each layer of the retina is determined using a novel variation of phasor analysis that is not limited to three components. The super-phasor unmixing (SPU) method is compared with fully constrained linear spectral unmixing. The seven spectroscopic signals in the spectral range 300-690 nm enable the quantitative unmixing of sub-micron pixel spectra even in the presence of noise. The performance of SPU was found to be significantly superior to linear unmixing especially in regions of high spectral overlap. The analysis being presented represents an important step in addressing chemical complexity in congested spaces with applicability to biomedical imaging and unmixing of hyperspectral images.