[Reliability of quantifying vascular white matter brain lesions -- a contribution to reproducible quantitative diagnosis].

PURPOSE

Microangiopathic lesions of the brain tissue correlate with the clinical diagnosis of vascular subcortical dementia. The "experience-based" evaluation is insufficient. Rating scales may contribute to reproducible quantification.

MATERIALS AND METHODS

In MRI studies of 10 patients, 9 neuroradiologists quantified vascular white matter lesions (WMLs) at two different points in time for 12 anatomically defined regions with respect to number, size and localization (score). For 9 observers and 10 studies, 90 intra-observer differences were obtained for each of the 12 WML scores. To calculate the inter-observer reliability, rating pairs were formed. Furthermore, 360 differences were computed for each score and rating for 12 anatomically defined WML scores, and the intraclass correlation (ICC) was calculated as a measure of agreement (reliability).

RESULTS

As to the intra-observer reliability, the median of the differences was 1.5 for the entire brain as opposed to 0 for defined brain regions. The corresponding values for the inter-observer reliability were 3 and 1, respectively. The mean intra-class correlation coefficient for the 10 studies was 0.88, whereas the mean interclass correlation concerning the inter-observer reliability was 0.70, with the first and second rating being averaged. The rating of each study took about 6 minutes.

CONCLUSION

The rating scale with high intra- and inter-observer reliability can dependably quantify WMLs and correlates with the clinical diagnosis of vascular dementia. Using a reliable rating scale, the diagnostic distinction of age-associated physiological vs. pathological size of the WML can make a contribution to the reproducible quantifiable diagnostic evaluation of vascular brain tissue lesions within the framework of dementia diagnostics.