Reproducibility and diagnostic performance of the vascular index of superb microvascular imaging in real-time breast ultrasonography for evaluating breast masses

Abstract

Purpose: This study aimed to evaluate the reproducibility and diagnostic performance of a quantitative parameter of superb microvascular imaging (SMI) in real-time breast ultrasonography (US) for differentiating benign from malignant breast masses. Methods: Eighty-seven breast masses in 75 patients who underwent both B-mode US and SMI before US-guided core needle biopsy were included in this study. Two radiologists performed B-mode US and measured the vascular index (VI) of SMI respectively for each lesion in real time. Intraobserver and interobserver agreements were analyzed for the VI of SMI. The diagnostic performance of B-mode US using the Breast Imaging Reporting and Database System lexicon and combined use with the VI of SMI was evaluated compared to pathology. Results: The median VI of malignant masses (n=32) was significantly higher than that of benign masses (n=55) (7.6% and 2.6%, respectively; P<0.001). The intraobserver agreement for VI was excellent regardless of the pathology, size, or depth of the lesion. The interobserver agreement for VI was excellent regardless of the presence of a measurement interval. The interobserver agreement for the final diagnostic decision was improved by combining B-mode US and VI (κ=0.883) in comparison with B-mode US only (κ=0.617). Adding VI led to significant improvements in the specificity (87.2% vs. 52.7%, 83.6% vs. 49.0%), accuracy (89.7% vs. 69.3%, 84.0% vs. 65.9%) and positive predictive value (81.5% vs. 55.1%, 75.6% vs. 52.6%) of B-mode US for both observers compared with B-mode US alone (all, P=0.001). Conclusion: The VI of SMI for real-time breast US is highly reproducible and leads to improved diagnostic performance for differentiating between benign and malignant breast lesions in combination with B-mode US.