Alzheimer Detection Using CNN and GAN Augmentation

Abstract

Alzheimer’s disease is a neurological condition which slowly weakens the memory, ability of thinking and reasoning along with the ability of performing day to day activities. The senior citizens are mostly affected with this disorder especially those in their sixties. The specific cause for this condition is still not clear but it can genetic, accidental or can be caused by some other circumstances considered as hypothesis at the moment. There are a few methods of detecting the disease but the MRI scans are the prominent ones among them. And for this research MRI scans are used in the form of scanned images. The disease has been classified into four categories for this research that are healthy, mild demented, very mild demented, and moderately demented. Deep learning algorithms are being used because of their efficient ways of working in the medical field. CNN the commonly used deep learning algorithm is kept as the base for the proposed model and the dataset is collected from Kaggle. The collected dataset is increased with the help of GAN augmentation to improve the accuracy of the model. The model gives accurate results up to 98.5% for detecting the disease and its categories. This model can help the medical workers in the form of a second opinion when combined with the present detecting techniques and can reduce their workloads.