The Interobserver Variability and Diagnostic Performance of 3-Dimensional Breast Ultrasound

• Published in: Ultrasonography (Seoul, Korea) Korea), 2011-09, Vol.30 (3), p.209
• Main Authors: Lyou, Chae Yeon, Kim, Sun Mi, Jang, Mijung, Kim, Sung Won, Kang, Eunyoung, Park, So Yeon, Moon, Woo Kyung
• Format: Article
• Language: English
• Subjects: Ultrasonic imaging
• ISSN: 2288-5919; 2288-5943

PURPOSE: We wanted to evaluate the interobserver variability and diagnostic performance of 3-dimensional (3D) breast ultrasound (US) as compared with that of 2-dimensional (2D) US. MATERIALS AND METHODS: We included 150 patients who received US-guided core biopsy and 3D US between June 2009 and April 2010. Three breast imaging radiologists analyzed the 2D and 3D US images using the Breast Imaging Reporting and Data System (BI-RADS) lexicon. The intra-observer agreement and inter-observer agreement were calculated. The sensitivity and specificity of 2D and 3D US were evaluated. RESULTS: The intra-observer agreement between 2D and 3D US was mostly slight or fair agreement. However, in terms of the final category, there was substantial agreement for all three radiologists. The inter-observer agreement of 3D US was similar to that of 2D US (moderate agreement for shape, orientation, circumscribed margin and boundary; fair agreement for indistinct margin, angular margin, microlobulated margin, echo pattern and final category). The sensitivity of 3D US for breast cancer was higher than that of 2D US for two radiologists (2D vs. 3D for reader 2: 55.8% vs. 61.5%, 2D vs. 3D for reader 3: 59.6% vs. 63.5%), and the specificity of 3D US was lower than that of 2D US for all the readers (2D vs. 3D for reader 1: 90.8% vs. 86.7%, 2D vs. 3D for reader 2: 90.8% vs. 87.8%, 2D vs. 3D for reader 3: 94.9% vs. 90.8%), but the difference was not significant (p > or = 0.05). CONCLUSION: The interobserver variability and diagnostic performance of 3D breast US were similar to those of 2D US.