Fingerprint of Amyloid Plaque in a Bitransgenic Animal Model of Alzheimer Disease Obtained by Statistical Unmixing Analysis of Hyperspectral Raman Data

Global prevalence of Alzheimer’s disease (AD) points to endemic levels, especially considering the increase on average life expectancy worldwide. AD diagnosis based on early biomarkers, in addition to a better knowledge of related pathophysiology are both crucial in the search of medical interventions able to modify AD progression. In this study we used unsupervised spectral unmixing statistical techniques to identify the vibrational spectral signature of amyloid β aggregation in neural tissue, as early biomarkers of AD in an animal model. We analyzed spectral images composed by a total of 55,051 Raman spectra obtained in frontal cortex and hippocampus from five bitransgenic APPswePS1ΔE9 mice, and colocalized amyloid β plaques by other fluorescence techniques. The Raman signatures provided multifrequency fingerprint consistent with the results of synthetized amyloid β fibrils. The fingerprint obtained from unmixed analysis in neural tissues are shown to provide detailed image of amyloid plaques in the brain, with potential to be used as biomarkers for non-invasive early diagnosis and pathophysiology studies in AD on retina.