Variability and validity of a simple visual rating scale in grading white matter changes on magnetic resonance imaging

Background and Purpose. A new simple visual rating scale can be used in magnetic resonance imaging (MRI) to grade the severity of white matter changes (WMC). The authors sought to study the interobserver variability and the validity of this visual rating scale against a computer-aided quantitative method in measuring WMC. Methods. The authors examined 220 magnetic resonance images from stroke-free patients with traditional risk factors for atherosclerosis and middle cerebral artery stenosis who were participants in a clinical trial requiring MRI examination. The severity of WMC was graded from 0 (no WMC) to 3 (severe WMC). For the quantitative assessment of WMC, the volume of WMC were measured with a computer-aided, automated segmentation method. The interobserver variability was also determined. Results. The mean volumes were significantly different among the different groups, which were scored 1, 2, and 3 (1324.59, 6475.48, and 25,440.05 mm(3), respectively; 1-way analysis of variance, P <.001). Interobserver variability for grading WMC was good in different locations of the brain. Excellent results were found in the frontal lobe (kappa = 0.829) and the parietal-occipital lobe (kappa = 0.646), whereas poor results were found in the basal ganglia (kappa = 0.391). There was perfect agreement between the 2 observers in 89.1% for the frontal lobe and 77.0% for the parietal-occipital lobe. Conclusions. The results suggest that this simple visual rating scale is a valid and reliable method for grading WMC in the lobar region but not in the basal ganglion.