Multispectral fluorescence imaging of atherosclerosis.

Multispectral fluorescence imaging is a new diagnostic technique with the potential to provide improved detection and classification of atherosclerotic disease. This technique involves imaging the fluorescence response of a tissue region through a tunable band-pass filtering device. The result is a set of images in which each individual image is composed of the fluorescence emission within a specified band of wavelengths. Multispectral imaging combined with angioscopic technology allows direct access to important spectral information and spatial attributes providing the potential for more informed clinical decisions about which, if any, treatment modality is indicated. In this dissertation, the system requirements for an angioscopic system with multispectral imaging capability are identified. This analysis includes a description of the necessary optical components and their characteristics as well as the experimental determination of spectral radiance values for the fluorescence response of human aorta specimens and the estimation of anticipated signal-to-noise ratios for the spectral images. Other issues investigated include the number of spectral images required to provide good classification potential and the best normalization method to be utilized. Finally, the potential utility of the information contained within a multispectral data set is demonstrated. Two methods of utilizing the multispectral data are presented. The first method involves generating a ratio-image from the ratio of the intensities of two spectrally filtered images. The second method consists of using histologically verified training data to train a projector and then applying that projector to a set of spectral images. The result (a weighted sum of the spectral images) provides improved contrast between normal and diseased tissue, and is called an optimized-contrast image. White-light images (generated using an incandescent light source), total-fluorescence images (the fluorescence response without spectral filtering), ratio-images, and optimized contrast images are compared. The results indicate that angioscopic fluorescence imaging appears to be a feasible and potentially useful technique in terms of providing improved detection of atherosclerotic disease. This technique warrants further investigation to further define the system requirements and to evaluate its clinical usefulness including the collection of a more extensive data set and the development of a prototype system