Study on Mild Cognitive Impairment and Alzheimer's Disease Classification Using a New Ontogenic Neural Architecture, the Supervised Reconfigurable Growing Neural Gas

Abstract

Alzheimer's Disease (AD) is one of the most prevalent aging-associated chronic diseases for the elderly population. Its prodromal stage is the Mild Cognitive Impairment (MCI). The detection of this stage versus AD is very difficult. We propose a new ontogenic neural architecture for dealing with the MCI-AD classification task. This is the Supervised Reconfigurable Growing Neural Gas (SupeRGNG), which is based on the Growing Neural Gas. We present a study on 495 Subjects from the Alzheimer's Disease Neuroimaging Initiative database, with 345 MCI and 150 AD. SupeRGNG yielded very good performance results just using six features related to neuropsychological tests: 0.98 accuracy, 0.98 specificity, 0.98 sensitivity, and 0.97 AUC. It outperformed many state-of-the-art proposals based on Deep Learning and neuroimaging. These findings suggest that our proposal may be an appropriate candidate for the early detection of AD in any clinical setting.