Deep Convolutional Neural Network based Classification of Alzheimer's Disease using MRI data

TL;DR

This paper presents a 2D deep convolutional neural network that accurately classifies MRI scans into Alzheimer's, mild cognitive impairment, and normal control with nearly 99.9% accuracy, improving over existing methods.

Contribution

The study introduces a novel 2D-DCNN approach for multi-class Alzheimer's diagnosis using imbalanced MRI data, achieving superior accuracy and robustness.

Findings

Achieved 99.89% classification accuracy.

Outperformed state-of-the-art methods.

Effective on imbalanced datasets.

Abstract

Alzheimer's disease (AD) is a progressive and incurable neurodegenerative disease which destroys brain cells and causes loss to patient's memory. An early detection can prevent the patient from further damage of the brain cells and hence avoid permanent memory loss. In past few years, various automatic tools and techniques have been proposed for diagnosis of AD. Several methods focus on fast, accurate and early detection of the disease to minimize the loss to patients mental health. Although machine learning and deep learning techniques have significantly improved medical imaging systems for AD by providing diagnostic performance close to human level. But the main problem faced during multi-class classification is the presence of highly correlated features in the brain structure. In this paper, we have proposed a smart and accurate way of diagnosing AD based on a two-dimensional deep convolutional neural network (2D-DCNN) using imbalanced three-dimensional MRI dataset. Experimental results on Alzheimer Disease Neuroimaging Initiative magnetic resonance imaging (MRI) dataset confirms that the proposed 2D-DCNN model is superior in terms of accuracy, efficiency, and robustness. The model classifies MRI into three categories: AD, mild cognitive impairment, and normal control: and has achieved 99.89% classification accuracy with imbalanced classes. The proposed model exhibits noticeable improvement in accuracy as compared to the state-fo-the-art methods.