Ensembles of Decision Trees Based Diagnosis of Alzheimer’s Disease and Dementia

Abstract

Efficient and precise diagnosis of Alzheimer’s disease (AD) and different types of dementia is a new topic of research. It is essential to have an accurate examination of both of these diseases for their medical treatment. However, the traditional classification has expensive, slow performance, and irreproducible. Therefore, our goal is to present an automatic computer-aided system for brain image classification. We use 66 T2-weighted images downloaded from Harvard Medical School. In this paper, we propose to use SMRI as a biomarker. First we use 2D discrete wavelet transform (2D-DWT) along with Daubechies-4 from wavelet families for feature extraction from MRI. Second, principal component analysis (PCA) is employed for the reduction of the feature to only 10. Third, those reduced feature were fed to Bagged Trees classifier for classification. A 6-fold cross validation (CV) has showed the classification accuracy of 69.7% .Thus, Bagged Trees classifier is effective and this approach can be a beneficial tool in practical.