ABSTRACT
Alzheimer's Disease (AD) remains a global health challenge which clearly calls for comprehensive approaches to help with its early detection. This research presents an innovative approach utilizing brain imaging data and advanced deep learning techniques to classify AD progression into four categories: Four types of individuals were included, namely: ‘MildDemented, ‘Moderate Demented,’ ‘NonDemented,’ and ‘VeryMildDemented.’ A new instance of CNN architecture that was modified was developed and trained with high accuracy of validation with 99% higher than standard mobile net that had 93% accuracy. The features extracted from the imaging data including the proposed cognitive features are used in the method to show better performance in discriminating between different stages of cognitive decline. The efficiency of the Modified CNN in early AD classification was further confirmed by other evaluation measurements such as accuracy, precision, recall, and F1-score. Thus, this work would complete a positive contribution to the field by providing an accurate assessment tool as early risk indicators and means of screening and management of patients at risk of AD could be initiated. The discoveries can be used to increase the efficacy of treatments with regard to this category of patients and provide the caregiving population with a higher quality of life.
