| Abstract: |
Our purpose was to evaluate Virtual Touch IQ (VTIQ) elastography and identify quantitative "rule-in" and "rule-out" thresholds for the probability of malignancy, which can help avoid unnecessary breast biopsies. 189 patients with 196 sonographically evident lesions were included in this retrospective, IRB-approved study. Quantitative VTIQ images of each lesion measuring the respective maximum Shear Wave Velocity (SWV) were obtained. Paired and unpaired, non-parametric statistics were applied for comparisons as appropriate. ROC-curve analysis was used to analyse the diagnostic performance of VTIQ and to specify "rule-in" and "rule-out" thresholds for the probability of malignancy. The standard of reference was either histopathology or follow-up stability for >24 months. 84 lesions were malignant and 112 benign. Median SWV of benign lesions was significantly lower than that of malignant lesions (p < 0.001). The application of a "rule-out" threshold of 1.9 m/s lead to a sensitivity of >98% with a concomitant significant (p = 0.032) reduction in false positive cases of almost 15%, whereas a "rule-in" threshold of 6.5 m/s suggested a probability of malignancy of >95%. In conclusion, VTIQ elastography accurately differentiates malignant from benign breast lesions. The application of quantitative "rule-in" and "rule-out" thresholds is feasible and allows reduction of unnecessary benign breast biopsies by almost 15%. © 2018 The Author(s). |
