Intra- and interobserver variability of linear and volumetric measurements of brain metastases using contrast-enhanced magnetic resonance imaging.

OBJECTIVES: To compare the intra- and interobserver variability of diameter and semiautomated volume measurements of brain metastases on contrast-enhanced magnetic resonance imaging (CE-MRI) data.
MATERIALS AND METHODS: About 75 MRI staging examinations of patients with metastasized renal cell carcinoma, thyroid cancer, or malignant melanoma (mean age, 56 years; range, 40-75 years) were included. Patients had been examined with a routine MRI protocol, including a CE 3D T1-weighted MP-RAGE sequence (1-mm slice thickness). MRI data were retrospectively analyzed using the OncoTREAT segmentation system (MeVis, Bremen, Germany, version 1.6). Volume of 355 enhancing brain metastases included in the analysis as well as the largest diameter according to Response Evaluation Criteria for Solid Tumors were measured by 2 radiologists. Intra- and interobserver variability was calculated.
RESULTS: Metastases (n = 355) had a mean diameter of 12.2 mm (range, 3.4-44.3 mm) and a mean volume of 1.4 cm(3) (range, 12-25.1 cm(3)). With respect to interobserver variability analysis revealed broader limits of agreement for response evaluation criteria for solid tumor measurements of all lesions (range, +/-27.8%-+/-33.0%; unsigned mean: 0.2%-2.5%) than for volume measurements (range, +/-21.4%-+/-23.3%; unsigned mean, 0.1%-0.3%) with statistically significant differences between diameter and volume measurements (P <or= 0.001). Limits of agreement were similar for intra- and interobserver comparisons.
CONCLUSION: Semiautomated segmentation of brain metastases on the basis of CE-MRI yielded reproducible volume measurements with a lower variability compared with linear measurements. Volumetry of contrast-enhancing brain metastases appears to be a suitable method for size determination in oncologic follow-up imaging.