The object of this proposal is to develop an automated method of generating spectrally resolved, topographical maps of histopathology samples. By collecting fluorescence, absorption, reflection and spectra it will be possible to positively identify the complete spectral signature from fluorescence stained tubercule bacilli or immunofluorescence stained biopsies from transplanted organs. Additional information on the physiology of tissue will be monitored and mapped by collecting the natural fluorescence of collagen, NADH, elastin, flavins and protoporphyrins. We propose to extract from these hyperspectral data sets the key changes in total spectral envelope of each component and identify the primary characteristics that separate one component from another. The proposed system comprises a research fluorescence microscope equipped with a unique imaging spectrometer coupled with a hybrid neural net data processor (HNN). The system automatically collects the image and identifies and maps the location on the sample of each hyperspectral signature identified as significant by a trained HNN. The entire microscope and software will be controlled by a PC. We expect this technique to be most effective in screening pulmonary specimens from tuberculosis patients and evaluate immunoperoxidase stained biopsies from patients who have undergone organ transplants and have clinical symptoms of humoral rejection.
An imaging spectrometer coupled to a fluorescence microscope will provide histopathologists with a sophisticated tool to aid in the rapid identification of abnormalities or bacteria. In fact, the micoscope is the worlds most commonly used inspection tool and the benefit of imaging spectral will add new power within a vast range of industrial applications including: semi conductor, plastics, fiber optics, and environmental monitoring.