Manuscript received February 19, 2025, revised April 7, 2025; accepted June5, 2025; published July 17, 2025.

Abstract—Alzheimer’s disease is a type of dementia that usually affects elderly people. It is a neurological disorder that causes a patient to lose memory gradually over time. The brain of an Alzheimer Disease (AD) patient shrinks due to the accumulation of amyloid plaques in the neurons. As a result, the neurons, which are the basic building blocks of the brain, lose connections and cannot communicate with each other. A person can be prevented from having AD if diagnosed at the right time. So, it’s very important to detect patients with mild symptoms of dementia to save them from getting AD. In this work, we have proposed a customized Convolutional Neural Network (CNN) model for classifying Alzheimer’s disease. The model has been evaluated with two benchmark datasets, the Kaggle Alzheimer’s dataset and the ADNI dataset. The two datasets differ in the number of images. The K-fold technique has been applied to overcome the problem of class imbalance. We have updated the model parameters using optimizers, namely Stochastic Gradient Descent (SGD), SGD with momentum, AdaGrad, AdaDelta, RMSprop, and Adam. Experimental results established that the proposed model outperforms many of the state-of-the-art models, considering the two benchmark datasets. In case of the Kaggle dataset, we have attained 99%accuracy using a customized CNN, outperforming other previous works that used a pre-trained model but still failed to produce 99% accuracy. Considering the number of images and class imbalance ADNI dataset also outperformed other previous models by achieving 90% accuracy. The main advantage of this work is that it studies the impact of all the state-of-theart optimizers with different epochs rather than experimenting with a particular optimizer and epoch. Optimizers have a huge impact on the performance of the model and also on the convergence time. It is an important hyperparameter that needs to be analysed further for better classification purposes.

Keywords—Alzheimer disease, CNN, class imbalance, K fold, optimizer

Cite: Pallavi Saikia and Sanjib Kr Kalita, "Performance Evaluation of Different Optimizers on Alzheimer’s Disease Classification Using a Customized Convolutional Neural Network," Journal of Image and Graphics, Vol. 13, No. 4, pp. 362-379, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.