Alzheimer’s Disease Classification Using Parallel Deep Convolutional Neural Network

Abstract

Alzheimer's disease (AD) is a neurological disorder that causes progressive cognitive decline in older adults. The relentless deterioration of AD symptoms makes it increasingly difficult for the patient to perform essential functions, leading to impaired immunity as well as accidents. Early diagnosis facilitates prompt therapeutic intervention to slow down symptoms and improve their quality of life. Convolutional neural networks (CNNs) can precisely examine brain imagery extracted from various imaging modalities. In this study, a parallel path deep CNN is employed to distinguish several early stages of AD from brain MRI scans. Parallel path CNNs offer a distinct advantage in image classification by simultaneously extracting features from multiple pathways, as opposed to single path CNNs that only operate with one pathway. The proposed network thus allows the integration of both local and global features, providing a comprehensive representation of the MRI images for an accurate classification of early AD stages. The Alzheimer's 4-class dataset, available on Kaggle, has been used to conduct the experiments for this work. The proposed model achieves 99.50% accuracy, 99.93% AUC, and a loss of 0.013 and outperforms the single path CNN models it was compared against. By effectively utilizing diverse features, our model overcomes the constraints associated with single path CNN architectures.