Evaluation of VGG-16 and VGG-19 Deep Learning Architecture for Classifying Dementia People

Dementia is a broad term that refers to a significant decline in one's ability to remember. Dementia is most commonly caused by Alzheimer’s, which is often difficult to diagnose and late. In fact, the very mild stage of dementia is the most effective stage of diagnosis. Therefore, it will be a massive advantage if the diagnosis is successful at an early stage. This paper attempts to evaluate the VGG-16 and VGG-19 architecture by appending a fully connected layer at the network's end to identify four classes of dementia: very mild dementia, mild dementia, and moderate dementia, as well as a non-dementia or normal people control class. The results of this paper successfully detect with an accuracy of up to 99%. The highest accuracy value was recorded at 99.68% for training and 99.38% for validation. The analyses include the value components of the confusion matrix, i.e., precision, recall, and F1 Score.