Development of Signal Processing Tools and Analysis of Hyperspectral Imaging Data in Diagnostics of Prion Diseases

Abstract

Creutzfeld-Jakob disease is a invariably deadly prion disease with no cure that attacks the brain. Hyperspectral microscopy has been used to examine 26 amyloid plaques (captured with two different filters) present in infected brains, stained with the luminescent polymer hFTAA. A MatLab program using the correlation coefficient between the emission spectrum from the sample and a reference spectrum representing the autofluorescence from the center of a plaque is used to examine where the staining is most evident. The correlation proved to be highest at the periphery of the plaques, indicating that the staining was most pronounced in the center. Two new programs were written to view the emission spectra for different distances from the center. The first program used pixels in the hyperspectral image lying on five circles with different radius, while the other used pixels from intensity based zones in the image. As it proved to be the most reliable, the latter was preferred used on the hyperspectral images in the data set. A distinct red shift in emission spectra as one moves from the periphery to the center of the plaques was revealed, as well as a strong increase of intensity and at around 606 nm.