Comparison of Hyper-Spectral Wide-Field and Confocal Fluorescence Microscopic Techniques

COMPARISION OF HYPER-SPECTRAL WIDE-FIELD AND CONFOCAL FLUORESCENCE MICROSCOPIC TECHNIQUES

Naga Srilakshmi Annamdevula¹, Ashley Stringfellow², Diego Alvarez^3,2, Thomas C.Rich² , Silas J. Leavesley^1,2

¹Chemical and Biomolecular Engineering, University of South Alabama, Mobile, AL,  ²Pharmacology, College  of Medicine, University of South Alabama, Mobile, AL,  ³Internal Medicine, College of Medicine, University of South Alabama, Mobile, AL

Identification and quantification of fluorescently-labeled cells in highly autofluorescent tissues is very difficult using single-wavelength fluorescence microscopy. We have previously used hyper-spectral imaging and linear unmixing analysis to detect GFP-expressing cells in autofluorescent tissues. In performing these studies, we have found significant variations in the sensitivity and specificity of GFP detection-dependent on the optical configuration and imaging detector. The goal of the study was to compare the effectiveness and sensitivity of these two spectral microscopy configurations (wide-field and confocal) for identifying fluorescently labeled cells in highly autofluorescent environment. We studied two different hyper-spectral fluorescence microscope systems: a wide-field system with an acoustic-optic tunable filter and electron-multiplying charge coupled device (EMCCD) camera and a laser scanning confocal microscope with a diffraction grating and 32-channel photo multiplier tube (PMT). These two systems differ in their sensitivity and signal collection, detector configuration, depth-of-field, and spectral filtering methods. To achieve this, we performed sensitivity analysis of multiple parameters on both systems to assess the effects of these parameters on detection sensitivity and specificity. Confocal pinhole, laser power, PMT gain, CCD detector gain, and rate of photobleaching were the independent parameters studied. Dependent variables included signal-to-noise ratio (SNR), root-mean-square error (RMS), and linear spectral unmixing analysis. Spectral image stacks were collected from both the systems. Image analysis was performed using MATLAB, Nikon Elements, and ENVI software packages. We have obtained preliminary results from the wide-field spectral fluorescence microscope. Spectral analysis is currently being performed on the spectral image stacks acquired using the confocal spectral microscope. As spectral imaging and analysis is a new approach to detect fluorescent cells in highly autofluorescent environment, this study will help researchers to optimize the parameters for a given hyper-spectral microscope system. These results should be applicable to many types of in-vitro and ex-vivo fluorescence assays.