An LDA and probability-based classifier for the diagnosis of Alzheimer's Disease from structural MRI

摘要

In this paper a custom classification algorithm basedudon linear discriminant analysis and probability-based weightsudis implemented and applied to the hippocampus measurementsudof structural magnetic resonance images from healthy subjectsudand Alzheimer’s Disease sufferers; and then attempts to diagnoseudthem as accurately as possible. The classifier works by classifyingudeach measurement of a hippocampal volume as healthy controlsizedudor Alzheimer’s Disease-sized, these new features are thenudweighted and used to classify the subject as a healthy controludor suffering from Alzheimer’s Disease. The preliminary resultsudobtained reach an accuracy of 85.8% and this is a similarudaccuracy to state-of-the-art methods such as a Naive Bayesudclassifier and a Support Vector Machine. An advantage of theudmethod proposed in this paper over the aforementioned state of the art udclassifiers is the descriptive ability of the classificationsudit produces. The descriptive model can be of great help to aid auddoctor in the diagnosis of Alzheimer’s Disease, or even further theudunderstand of how Alzheimer’s Disease affects the hippocampus.