Deconvolution of Skin Images with Multivariate Curve Resolution

Two-photon fluorescence microscopy has been demonstrated as a high-resolution 3-D tissue imaging technique. If the dream of a noninvasive optical biopsy is to be realized upon this technology, biochemically relevant information must be extracted from the data thus obtained. The current study investigated the ability of different Gram-Schmidt modifications to orthogonal projection multivariate curve resolution (MCR) to find the emission spectra and distributions of fluorophores in volume images. A two-photon microscope, incorporating a high-power femtosecond Ti:Sapphire laser and a high sensitivity multianode photomultiplier spectrally resolved detector, was used to obtain spectroscopic data for each point in a volume of the sample. The MCR algorithm was developed and tested by first examining fluorescent beads. MCR analysis on images obtained from ex vivo human skin provided recognizable structural resolution with no a priori information, suggesting that the resolved spectra correspond to different endogenous fluorophores.