A new model for early diagnosis of alzheimer's disease based on BAT-SVM classifier

Shereen A Taie,Wafaa Ghonaim

doi:10.11591/eei.v10i2.2714

Abstract

Magnetic Resonance Images (MRI) of the Brain is a significant tool to diagnosis Alzheimer's disease due to its ability to measure regional changes in the brain that reflect disease progression to detect early stages of the disease. In this paper, we propose a new model that adopts Bat for parameter optimization problem of Support vector machine (SVM) to diagnose Alzheimer’s disease via MRI biomedical image. The proposed model uses MRI for biomedical image classification to diagnose three classes; normal controls (NC), mild cognitive impairment (MCI) and Alzheimer’s disease (AD). The proposed model based on segmentation for the most involved areas in the disease hippocampus, the features of MRI brain images are extracted to build feature vector of the brain, then extracting the most significant features in neuroimaging to reduce the high dimensional space of MRI images to lower dimensional subspace, and submitted to machine learning classification technique. Moreover, the model is applied on different datasets to validate the efficiency which show that the new Bat-SVM model can yield promising acceptable level of accuracy reached to 95.36 % using maximum number of bats equal to 50 and number of generation equal to 10.

Highlights

Alzheimer disease is considered as an evolution neurodegenerative disease, the conditions that result in the loss of memory and cognitive abilities which represent a public health problem
Dataset The dataset that has been used for training and testing the proposed system was obtained from the Radiopaedia.org
The goal of Radiopaedia to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer's disease (AD)

Summary

INTRODUCTION

Alzheimer disease is considered as an evolution neurodegenerative disease, the conditions that result in the loss of memory and cognitive abilities which represent a public health problem. Numerous MRI techniques were proposed as non-invasive imaging biomarkers for the early diagnosis of AD and evaluation of disease progression. These biomarkers were analyzed to study of differences between groups, such as disease and healthy groups. These biomarkers methods can not improve the clinical potential diagnosis due to these are not applicable on a single-subject level To overcome these limitations machine-learning techniques were promising tools in data analysis of neuro imaging for individual class prediction. MRI data have a brain imaging powerful analysis tool which is machine learning techniques that aimed to classify AD vs cognitively normal (NC) or MCI vs NC [4]. A new Bat-SVM model was presented for optimizing the parameter of the problem of the SVM to maximize the classification accuracy reached to 95.36%

PROPOSED MODEL

EXPERIMENTAL RESULTS AND DISCUSSION

CONCLUSION