Abstract
Early detection of Alzheimer's disease is expected to aid in the development and monitoring of more effective treatments. Classification methods have been proposed to distinguish Alzheimer's patients from normal controls using Magnetic Resonance Images. However, their performance drops when classifying patients at a prodromal stage, such as in Mild Cognitive Impairment. Most often, the features used in these classification tasks are related to structural measures such as volume, shape and tissue density. However, microstructural changes have been shown to arise even earlier than these larger-scale alterations. Taking this into account, we propose the use of local statistical texture maps that make no assumptions regarding the location of the affected brain regions. Each voxel contains texture information from its local neighborhood and is used as a feature in the classification of normal controls and Mild Cognitive Impairment patients. The proposed approach obtained an accuracy of 87% (sensitivity 85%, specificity 95%) with Support Vector Machines, outperforming the 63% achieved by the local gray matter density feature.