OBJECTIVE: To compare brain structures using volumetric magnetic resonance imaging with isotropic resolution, in T1-weighted gradient-echo (GRE) acquisition, with and without inversion recovery (IR). MATERIALS AND METHODS: From 30 individuals, we evaluated 120 blocks of images of the left and right cerebral hemispheres being acquired by T1 GRE and by T1 IR GRE. On the basis of the Naidich et al. method for localization of anatomical landmarks, 27 anatomical structures were divided into two categories: identifiable and inconclusive. Those two categories were used in the analyses of repeatability (intraobserver agreement) and reproducibility (interobserver agreement). McNemar's test was used in order to compare the T1 GRE and T1 IR GRE techniques. RESULTS: There was good agreement in the intraobserver and interobserver analyses (mean kappa > 0.60). McNemar's test showed that the frequency of identifiable anatomical landmarks was slightly higher when the T1 IR GRE technique was employed than when the T1 GRE technique was employed. The difference between the two techniques was statistically significant. CONCLUSION: In the identification of anatomical landmarks, the T1 IR GRE technique appears to perform slightly better than does the T1 GRE technique.

Keywords: Magnetic resonance imaging; Brain/anatomy & histology; Reproducibility of results.