The diagnosis of alzheimer's disease based on the classification of fMRI data

Alzheimer's Disease is the most common form of dementia. This thesis aims to use functional magnetic resonance imaging data classification to diagnose the disease. One of the main causes for Alzheimer's Disease is synaptic failure. An fMRI experiment that targets the brain regions that are affected by this failure has been developed. Differences are expected to emerge between the subjects in different groups. Five novel methods has been proposed to locate the activations. Three of these can detect an activation instantaneously, while the other two use the complete experiment data. The methods use robust regression to find the regression line for the resting state and use the distance to this line as a new metric. Besides activation information, functional connectivity methods has been researched which can be used to detect the Alzheimer's Disease. When the brain is at rest, the temporal correlations between specific regions form a network called the default mode network. There is a decrease in correlation in subjects with Alzheimer's Disease. The activations and functional connectivity features has been classified with LVQ and artificial neural networks. In the three data sets used, a success of 98% has been reached for the activation features, and 90% has been reached for functional connectivity.