An Attention-based CNN Architecture for Alzheimer’s Classification and Detection

Medical professionals employ machine learning extensively because it enables them to identify and categorize a wide range of fatal diseases for which there is no known treatment but can be avoided with early detection. Examples of such diseases include cardiovascular illness, brain tumors, kidney-related disease, Alzheimer’s disease, and many more. Alzheimer’s disease is the main factor contributing to dementia. Dementia impairs people’s capacity for autonomous functioning by reducing their reasoning and interpersonal coping capacity. In the beginning, the patient will forget previous occurrences. As the condition worsens, individuals will gradually lose the memory of entire incidents. The disease must be identified as soon as feasible. Using a machine learning approach, the MRI images from the Alzheimer’s Disease Neuroimaging (ADNI) database are classified into four classes of dementia Non-Demented (normal), Early-Mild Dementia, Mild Dementia, and Late Mild Dementia. In this study, we propose to use CNN as they are best used for Image detection and to improve the efficiency of the model Attention mechanism is applied. The accuracy shown by the CNN model is around 95% and using the Attention mechanism accuracy is coming to be 98.54