An evaluation of the significance of areas of intense signal in the MR brain images of patients with multiple sclerosis.

This study evaluates the distribution of areas of intense signal (AIS) in 189 patients with multiple sclerosis (MS) and 83 patients presenting high-risk factors of cerebrovascular injuries. Two multivariate statistical analyses (multifactorial discriminant analysis and logistic regression analysis) with two AIS scores and several subpopulations of patients (according to age and/or the number of AIS) were tested. The results of these analyses were expressed with the usual screening test. The results obtained in this study even without the help of any clinical information are very promising, since they established that the specificity of MRI could be improved by using the distribution of the AIS in the various anatomical areas as a criterion. Five regions of the brain display were particularly significant in the discrimination between MS and non-MS patients: In decreasing order we found that the temporal, occipital, brain stem, and parietal regions were more specific with respect to the diagnosis of MS, while only the basal ganglia could account for non-MS patients. With multifactorial discriminant analysis and logistic regression analysis respectively, 78.9 (+/- 2.8)% and 85.1 (+/- 2.8)% of the patients were correctly classified by MRI. The results obtained on the main group were confirmed by a predictive test carried out on an other population of 40 patients, which produced similar results. The comparison between our method and Fazekas's imaging criteria showed a 20% improvement in favour of our approach and it is hoped that it will contribute to make the most of MRI as a tool for the diagnosis of MS.