Multi-classification of alzheimer disease on magnetic resonance images (MRI) using deep convolutional neural network (DCNN) approaches

Alzheimer s disease (AD) is the most popular cause of dementia. Dementia refers to a continuous decline in mental ability. The developmental stages of this neuropsychiatric symptom are usually examined using medical images of the brain. Cutting edge technologies, including computer algorithms have been applied to the diagnosis and treatments of AD, especially in the area of detection and classification. These technologies have eased up the work of medical experts and provided faster ways of medical delivery. This study was aimed at advancing AD image classification with deep convolutional neural network (DCNN) involving convolutional neural network (CNN) and transfer learning (Visual Geometry Group (VGG)16 and VGG19) using magnetic resonance images (MRI) and extend the evaluation metrics since limitations and capacity of algorithms cannot be revealed by few metrics. The objectives of this research are to classify AD images into four known classes by neurologists and results of the finding are subjected to many evaluation metrics. This research used computer algorithms majorly DCNN and indeed transfer learning to classify AD. Six metrics were used for the evaluation; accuracy, area under curve (AUC), F1- score, precision, recall and computational time. VGG-19 was the best in three, CNN was the best in two and VGG-16 was the best in one. Conclusively, this study has proven that computer algorithms are capable of classifying AD into four classes known to medical experts.