Early Prediction and Analysis of DTI and MRI-Based Alzheimer’s disease through Machine Learning Techniques

Abstract

Alzheimer’s disease (AD) is a neurodegenerative disease generally occurring in 65 years or older, destroying neurons and various brain areas. Initially mild, the symptoms develop increasingly severe over time. Patients with AD are becoming more numerous every day. As a result, it is essential to detect AD progression early. Different clinical methods and neuroimaging techniques are used to detect this disease. Due to the complicated nature of AD, only clinical methods or neuroimaging techniques cannot correctly detect early AD and the progression of MCI patients. Besides, these techniques are costly, time-consuming, and limited availability. This research work uses Alzheimer’s disease Neuroimaging Initiative (ADNI) database to make significant predictions. Numerous machine learning models were examined to recognize early AD and mild cognitive impairment in cognitively normal people with identical features. Gaussian Naive Bayes identifies Alzheimer’s patient’s mild cognitive impairment and healthy people with a better classification accuracy of 96.92% using the selected and correlated features—ADMCI3, AV45, APOE4, AV45AB12, ADASQ4, RAVLT_perc_forgetting, AD_CGH_L, MD_CGH_L, RD_CGH_L, etc. than other models. The study findings showed that by using neuropsychological data combined with cognitive data, machine learning techniques could help diagnose AD.