Alzheimer’s Disease Detection from Brain MRI Data using Deep Learning Techniques

Alzheimer disease(AD) is a neurological disorder. For the AD, there is no specific treatment. Early detection of Alzheimer’s disease can help patients receive the correct care. Many studies employ statistical and machine learning techniques to diagnose AD. The human-level performance of Deep Learning algorithms has been effectively shown in different disciplines. In the proposed methodology, the MRI data is used to identify the AD and Deep Learning techniques are used to classify the present disease stage. For the classification and prediction of AD, we have constructed CNN architectures employing transfer learning. DenseNet121, MobileNet, InceptionV3 and Xception neural networks are trained using Kaggel AD dataset. All models in this study are trained on the same dataset in order to analyse their performances. The DenseNet121 architecture gives the highest accuracy of 91% on the test data that detects AD accurately.