Classification Of Alzheimer’s Disease Using Pre-Trained Convolutional Neural Networks Techniques

Abstract

The most prevalent and common type of dementia is Alzheimer's disease. It is remarkable; however, that very few people with Alzheimer Disease are diagnosed accurately and quickly. There is still no definitive cause or cure for the disease. Symptoms may be more manageable and their treatment may be more effective when the impairment is still at an early stage or MCI (Mild Cognitive Impairment) stage. In this work MRI images were taken. MRI scans are manually analyzed by radiologists to detect abnormal conditions in the brain. It takes a long time and it is difficult to manually interpret a large number of photos. However, the complexity associated with the MRI system makes this task non-trivial. Especially diagnosing the stages of Alzheimer into normal, mild, very mild, and moderate is not too easy and which tends to errors. To address the issue, computer aided diagnosis helps in fast, accurate and exact diagnosis of the disease. In the proposed work, Xception and Mobilenetv1 were used. Initially Xception and Mobilenetv1 network model all the layers were trained, the dense layer is added with the softmax classifier which classifies the Alzheimer Disease into normal, mild, very mild, and moderate. The comparison of performance the Xception model perform with higher accuracy 92.15%.