Abstract
Dementia, a prevalent neurodegenerative condition, is a significant aspect of Alzheimer's disease (AD), causing progressive impairment in daily functioning. Accurate and timely classification of AD stages is crucial for effective management. Machine learning and deep learning models have shown promise in this domain. In this study, we proposed an approach that utilizes support vector machine (SVM), random forest (RF), and convolutional neural network (CNN) algorithms to classify the four stages of dementia. We augmented these algorithms with watershed segmentation to extract meaningful features from MRI images. Notably, our findings demonstrate that SVM with watershed features achieves an impressive accuracy of 96.25%, outperforming other classification methods. The evaluation of our method was performed using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, and the inclusion of watershed segmentation was found to significantly enhance the performance of the models. This study highlights the potential of incorporating watershed segmentation into machine learning-based AD classification systems, offering a promising avenue for accurate diagnosis and treatment planning.