The Influence of Training on Breast Ultrasound Expertise

This study was designed to evaluate the effectiveness of the training experience to develop expertise for breast ultrasound. The same group of radiology residents was evaluated before and after training on the use of the Breast Imaging Reporting and Data System (BI-RADS) to develop breast imaging expertise. An evaluation set of breast ultrasonography images of pathologically confirmed lesions that consisted of 100 masses (54 benign and 46 malignant masses) without clinical information was developed. Two experienced breast radiologists independently reviewed cases to reach a consensus. Six radiology residents (four second-year residents and two third-year residents) trained one month to interpret breast ultrasound examinations, and to read and to review teaching files with expertise. The residents evaluated the examination set according to BI-RADS for shape, orientation, margin, lesion-boundary, echo pattern, posterior echo features, surrounding tissue, calcification and final assessment before and after the training. Training performance was measured by means of agreement (k) and diagnostic performance. After training, the second-year residents showed statistically significant increased agreement as compared to the experts for 15 of the total 36 BI-RADS descriptors. The third-year residents showed increased agreement for three of the 18 descriptors. The mean sensitivity for the residents was 94.2% (range, 89.1-100%) before training and the mean sensitivity increased to 97.8% (range, 95.7-100%) after training. The mean specificity was 50.6% (range, 40.7-59.3%) before training and the mean specificity decreased to 42.0% (range, 31.5-51.9%) after training. In this study, the use of breast ultrasonography training improved the ability of residents to determine lesion description, final assessment and coherence