Applied Machine Learning to Identify Alzheimer's Disease through the Analysis of Magnetic Resonance Imaging

Abstract

Alzheimer's disease is among the most common neurodegenerative diseases [1], doubling the number of patients every 5-year interval beyond age 65 [2]. Different investigations have proven that patients with Alzheimer's disease, show volume reduction at specific areas of the brain [1, 3-11]. Some of these areas, like the precuneus, start showing atrophy since early stages of the disease [1, 3, 6, 12-14], as measured through the use of Magnetic Resonance Imaging [9]. Considering this, we studied the possible use of the precuneus as a biomarker to identify such disease. Our results suggest that the precuneus is a potential biomarker to detect Alzheimer's disease, since 7 out of 10 patients (73.33% of accuracy) can be correctly classified.