Spatial regularization for neural network and application in Alzheimer's disease classification

Since the computer hardware power has been improved, neural networks have been widely applied in different fields. However, to the best of our knowledge, the idea of coding spatial information for Alzheimer's disease classification in the regularization neural network context is still new. This motivation has led us to propose a spatial regularization approach for neural network and apply it to structural MRI Alzheimer's disease classification. We tested our proposed model on two types of data. Real application results show that spatial regularization neural network increase the classification accuracy.