Ultrasound BI-RADS classification and real-time elastography Tsukuba score assessment of breast lesions: inter-and intraobserver agreement

Introduction: Breast ultrasonography is a widely recognized diagnostic tool for evaluating breast lesions. Breast lesions are classified according to the Breast Imaging Reporting and Data System (BI-RADS) in categories 2 – 5. Alternatively, real time elastography (RTE) explores stiffness differences between breast lesions and the surrounding normal parenchyma at the macroscopic scale and is classified in a Tsukuba elasticity score (TS). Our aim was to prospectively evaluate inter-and intraobserver agreement of BI-RADS classification and TS of breast lesions.

Material and methods: Women with breast lesion who were scheduled for ultrasound-guided invasive breast biopsy at the Women's Hospital of the University Hospital Basel, from August 2009 – December 2012 were recruited. The study included 164 breast lesions (63 malignant, 101 benign) from 156 patients. From each lesion, one representative B-mode image and 1 to 5 elastograms were recorded. Lesions were assessed by US BI-RADS classification and TS by the primary examiner. Three experienced reviewers independently assessed the B-mode image and the elastograms and repeated the evaluation after 2 months.

Results: Weighted kappa values for interobserver agreement ranged from moderate to substantial for BI-RADS classification (κ= 0.585 – 0.738) and was substantial for TS (κ= 0.608 – 0.779). Intraobserver agreement was almost perfect for US BI-RADS (κ= 0.847 – 0.872) and TS (κ= 0.879 – 0.914). Individual reviewers are highly self-consistent (almost perfect intraobserver agreement) with regard to BI-RADS classification and TS, whereas interobserver agreement was moderate to substantial.

Conclusion: Our results indicate that a high diagnostic performance of breast US and RTE can be achieved and both methods should be combinded for better evaluation.