Optical imaging methods have become increasingly popular in biomedical applications, due in part to recent advances in detector technology, increased availability, and decreased pricing. In vivo optical imaging techniques are especially promising, as these technologies do not bear the high cost or radiation concerns associated with more traditional imaging methods, such as X-ray, computed tomography (CT), or magnetic resonance imaging (MRI). However, bulk tissue optical properties generally have high scattering and attenuation coefficients that limit the penetration depth, sensitivity, and resolution of optical imaging methods. If fluorescence targeting methods are applied, tissue autofluorescence (background) also serves to limit imaging sensitivity. It is highly important to research methods for increasing the sensitivity to fluorescence signals in tissue, as targeting of specific cellular phenotypes often occurs at pmol or smaller concentrations. This is especially important in detection of rare cellular events, such as some cancers and other pathologies.

In this research, we have used a combination of hyperspectral imaging and mathematical processing techniques to isolate a fluorescence signal from autofluorescence background. We have designed a hyperspectral excitation-scanning imager that has been used to image multiple fluorescence signals in mouse models in the presence of high autofluorescence background. Visualization, analysis, and classification software has also been designed to allow separation of fluorescence signal from autofluorescence background and discrimination amongst multiple fluorescence signals of interest. This presentation will include a summary of the current hyperspectral imager design, optimization and construction as well as a description of the visualization, analysis, and classification methods implemented. Finally, a discussion of current applications and possibilities for translation of hyperspectral technology to the clinical imaging arena will be presented.