Ensemble of CNN Models for Identifying Stages of Alzheimer's Disease: An Approach Using MRI Scans and SMOTE Algorithm
Zihuan Li, Yinsong Wang, Zhuxin Jiang, Zitai Luo, Junlong Wu, Teoh Teik Toe
Abstract
This paper proposes an ensemble of convolutional neural network (CNN) models for identifying different stages of Alzheimer's disease (AD). The dataset used in this study, sourced from the Kaggle repository, comprises 6400 MRI brain scans of AD patients classified into four stages based on disease severity: “Mild Demented,” “Moderate Demented,” “Non Demented,” and “Very Mild Demented.” Prior to analysis, the dataset was preprocessed by cropping out the region of interest (ROI) and normalizing the images. To overcome the issue of imbalanced data, the Synthetic Minority Over-sampling Technique (SMOTE) algorithm was employed. The study highlights that SMOTE improves the efficiency of medical image analysis. Subsequently, a CNN-based approach was used for classification, and the model's performance was evaluated. The results demonstrate that the ensemble model outperforms individual CNN models in terms of accuracy. This research contributes to the advancement of neuroimaging and clinical analysis methods for predicting AD, and it demonstrates the efficacy of using an ensemble of CNN models to classify different stages of the disease.