Abstract

The escalating prevalence and economic burden of dementia underscore the urgency for innovative detection methods. This study investigates the potential of hyperspectral imaging (HSI) to detect dementia by analyzing retinal changes associated with amyloid beta (Aβ) formations. Leveraging a dataset of 3,256 ophthalmoscopic images from 137 participants aged 60 to 85 years, categorized into dementia and non-dementia groups via the Mini-Mental State Examination (MMSE), we extracted features from five key regions of interest (ROIs) identified for their pronounced changes in Aβ biomarkers. The analysis revealed that gender does not significantly influence dementia levels, and no substantial spectral differences were observed within the 380 nm to 600 nm wavelength range. However, significant variations in spectral reflection intensity were noted between 600 nm and 780 nm across both genders, suggesting a potential avenue for distinguishing stages of dementia. Despite the impact of diabetes on the vascular system, its stages did not significantly influence dementia development. This research highlights the utility of HSI in identifying dementia-related retinal changes and calls for further exploration into its effectiveness as a diagnostic tool, potentially offering a non-invasive method for early detection of dementia.