Longitudinal follow-up of individual white matter hyperintensities in a large cohort of elderly

We report on a method for the longitudinal follow-up of individual white matter hypersignals (WMH) and on its application to the study of WMH natural evolution in a cohort of 1,118 elderly over a 4-year period. For each subject, automated WMH detection was performed on T2-weighted MR images acquired both at baseline and at follow-up after registration in a common space. The detection algorithm was designed both to track WMH previously existing at baseline and to identify newly formed WMH. The average annual change in WMH load was found to be 0.25 cm(3)/year, 36% of this change being attributable to newly formed WMH. Quantitative analyses showed that change in WMH was mainly explained by progression of juxtaventricular and periventricular WMH while the load of WMH in the deep white matter zones was found stable over 4 years of the study. Statistical parametric mapping confirmed these spatial WMH change distributions in the juxta- and periventricular zones. High blood pressure was not a significant predictor of the annual change in WMH. This study proposes a new scheme for the longitudinal study of WMH change by dissociating worsening of existent WMH from surfacing of new WMH and may thus contribute to help understanding and characterizing the neurological and etiological bases of these two processes and their potential differences.