Alzheimer’s Disease Classification Using Feed Forwarded Deep Neural Networks for Brain MRI Images

The rise of dementia among the old population across the world will rapidly make financial suffering on healthcare industries, yet convenient acknowledgment of early notice for dementia and appropriate reactions to the event of dementia can upgrade clinical treatment. Usage of medical service data and health behavior are generally more available than clinical information, and a pre-screening apparatus with effectively open information could be a decent answer for dementia-related issues. In this chapter, we applied different deep neural networks (DNN) algorithms including Convolutional Neural Networks (CNN), Residual Neural Networks (RNN), Inception V3, and Dense Neural Networks (Densenet) were applied to the classification of MRI brain images. We considered brain images of 1098 subjects data collected from OASIS-3 imaging datasets whose age range was between 42 and 95. The system has been run with and without fine-tuning of features. The comparison of different models was performed and it is found that CNN and Dense net was outperformed other models and provided comprehensive performance outcomes with an accuracy of 95.7%, and 95.5%, respectively. This method can help both patients and doctors on early pre-screening of possible dementia.