Automated Alzheimer's Disease Stage Prediction Using Deep Learning Combined Approach

Abstract

Alzheimer's disease is a Declining neurological illness that mainly affects memory, thinking, and behaviour, leading to Significant deterioration in intellectual functioning of independence in daily routine. It impacts about 55 million people around the globe, with the number expected to rise significantly as the global population ages, making it one of the biggest public health challenges in current time. This study used a multi-case design to develop and test a deep learning model for classifying Alzheimer's disease stages from MRI scans. Two different datasets were used, the first one was used for the training and the other one was used for the validation task. Both of them used about 15000 images for the training and testing. A combination of VGG16 and LSTM neural networks was implemented using Keras and TensorFlow, with the VGG16 extracting visual features from MRI sequences which were then fed to an LSTM layer to classify sequences based on both visual and temporal dependencies. The model achieved 98.88% accuracy on the training set and 98.03% on the held-out validation set of 8,000 MRI images for classifying Alzheimer's disease stages. The combined VGG16-LSTM deep learning architecture demonstrated high performance in classifying Alzheimer's disease stages from MRI data, achieving over 98% accuracy. This research presents a promising automatic approach for diagnosis of Alzheimer's that has potential for clinical relevance