Abstract
Background: Utilization of multidetector-row CT (MDCT) is anticipated to improve the diagnostic accuracy and reliability for determining malignant cysts.
Purpose: To assess the diagnostic accuracy, interobserver agreement, benefit of consensus reading, and strong predictors of malignancy in determining malignant cystic renal masses at MDCT.
Material and Methods: Two radiologists independently rated the probability of malignancy at MDCT in 72 benign and 53 malignant cysts. The accuracy and interobserver agreement for determining malignant cysts were evaluated. The strong predictors of malignancy were determined, and in patients with interobserver disagreement for determining malignant cysts, consensus readings were performed.
Results: Az value of the two readers was 0.905–0.936 and the sensitivity and specificity were 85–89% and 83–93%, respectively. The overall interobserver agreement for determining the malignant cyst was good as the κ value was 0.696 (% agreement, 61% (76/125)). Thickened irregular wall, thickened irregular septa, and enhancing soft tissue component were strong predictors for malignancy with both readers. In the 17 patients with interobserver disagreement for determining malignant cysts, the sensitivity was improved from 38–63% to 89% by the consensus reading.
Conclusion: At MDCT, some false negative decisions for determining malignant cysts can be corrected by consensus reading, and thickened irregular septa, thickened irregular wall, and enhancing soft tissue component are the strong predictors of malignant cysts.
Acknowledgement
This work was supported by the Korea Research Foundation Grant funded by a grant (2009-639) from the Asan Institute for Life Sciences, Seoul, Korea.
Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.