Characterization of atypical cystic renal masses with MDCT: comparison of 5-mm axial images and thin multiplanar reconstructed images.

OBJECTIVE

The purpose of this study was to investigate whether cystic renal masses are better characterized on thin axial and multiplanar reconstructed MDCT images than on 5-mm images.

MATERIALS AND

METHODS

The records of 70 complex cystic renal masses in 59 patients (45 men, 14 women; mean age, 68 +/- 13 years) who underwent 64-MDCT at two medical centers were studied. Twenty-three of the masses were confirmed on the basis of the histologic findings and 47 in 2-4 years of follow-up. Images were reviewed in two sessions by two radiologists with 12 and 2 years of experience. In the first session, 5-mm axial images were analyzed, and in the second, thin axial images and multiplanar reconstructions. To assess intraobserver variability, analysis was repeated after 1 month. Statistical analysis was performed with Wilcoxon's signed rank test, receiver operating characteristic analysis, and weighted kappa statistics.

RESULTS

Radiologists 1 and 2 detected thicker cystic walls (p < 0.001, p < 0.005) and septa (p < 0.03, p < 0.05) and fewer septa (p < 0.005, p < 0.002) on 5-mm axial images and assigned significantly different Bosniak categories than they did in analysis of the volume data (p < 0.04, p < 0.05). Variability was reduced in thin axial and multiplanar views. No significant differences were found in characterization of lesions as benign or malignant in review of 5-mm axial images and volume data sets. The areas under the receiver operating characteristic curve were 0.89 for 5-mm images and 0.96 for volume data sets for radiologist 1 and 0.87 and 0.90 for radiologist 2.

CONCLUSION

Analysis of volume data sets is associated with less intraobserver and interobserver variability than review of 5-mm axial images. Wall thickness and the number and thickness of septa may differ, resulting in assignment of different Bosniak categories. Diagnostic performance in characterizing lesions as benign or malignant, however, is not statistically different for the thick and thin images.