Detection and classification of MS using magnetisation transfer ratio images

Principal component (PCA) and multiple discriminant analysis (MDA) are applied to magnetization transfer ratio (MTR) images in multiple sclerosis (MS). PCA and MDA are used to characterise subtle diffuse changes in MS. PCA is applied to MTR histograms to identify regions of significant variation. These areas are indicated as possible lesion areas. We compare two classifiers to recognise differences between normal controls and different types of MS disease; a Bayesian classifier is trained in PC space, and the histogram space is transformed to the optimal discriminant space for a nearest neighbor classifier.