Investigating the Impact of Different Data Representation with Several Classification Models on Magnetic Resonance Imaging (MRI)

Abstract

Alzheimer is one of the dementia diseases. It often infects elderly people who are beyond 65 years old. Alzheimer diagnosis (AD) relies on the analysis of the magnetic resonance imaging (MRI) and the scale of the clinical dementia rating (CDR). The automatic AD has widely been carried out adopting different methods. The investigation of the digital MRI samples with CDR scale has been the cornerstone of the automatic studies of AD. Those studies have resulted in various accuracy rates when they applied diverse techniques. This paper presents and compares several techniques of image normalization and classification models for improving the accuracy rates of AD. Four normalizations techniques, which is Z-score, Min-Max, Decimal Scaling and Standard deviation are examined. Whilst, the classifiers are Naïve Bayes, Logistic regression, decision tree (DT), k-nearest neighbors (kNN), Artificial Neural Network (ANN), and support vector machine (SVM). The analyzed dataset involves MRI from 150 individuals and its CDR scales. The experiments of this paper has been implemented using Orange software. Although the overall accuracy rates are quite good, the best findings are 91.2% from logistic regression when it analyzed the normalized dataset using Z-score. Comparing with other studies, the results of this study have shown significant improvement based on utilizing the normalization techniques.