Chapter 7 - Dementia detection using the deep convolution neural network method

Abstract

Dementia is a persistent and dynamic intellectual decline of mental capacity due to weakness. Accurate identification of the problem and timely diagnosis are the most crucial aspects of dementia. To handle this problem, neuroimaging with computer-aided mechanisms has gained significant advances in recent years. A particular variety of dementia is the next most regular sort of gradually increasing disease that arises with Alzheimer's. It is due to abnormal deposits of proteins in and around the cells of the brain. An example protein is amyloid, which generates plaques around the brain cells. Another protein labeled tau produces tangles in the brain cells. The indications of dementia are problems related to memory decline, increasing confusion, reduced concentration, changes in behavior, etc. Although there is an improvement in diagnosis, the ability to recognize dementia in clinical practice remains low. Neuroimaging with computer-aided algorithms has made striking advances in the discovery of dementia. Analysis of neuroimages by utilizing machine learning and deep learning mechanisms has promising outcomes in the detection and diagnosis of dementia. This chapter explores a model based on deep convolution neural networks to identify dementia using magnetic resonance imaging (MRI) scan images. The pretrained model, Inception-V3, is retrained for that purpose. Experimentation is performed by considering the brain MRI dataset called OASIS-1. The retrained network model achieves excellent results in detecting whether a person has dementia or not.