Implementation of an automated breast ultrasound system in an academic radiology department: Lesson learned in the first three years

This is the authors' abstract. We don't add key points for this paper.

Purpose: To evaluate the diagnostic performance of an ABUS in an academic radiology department over the first three years after its implementation. Methods: In this retrospective study women undergoing ABUS examination for screening and diagnostic purposes between October 2015-2018 were included in case of sufficient follow-up and established diagnosis. Women underwent ABUS + /- mammography in the same day. BI-RADS 1/ 2 cases with cancer diagnosis during follow-up and already visible in the previous exam were considered false negative (FN). BI-RADS 3/4 cases proved benign were considered false positive (FP). FP and number of additional targeted HHUS (addHHUS) were compared over the three years. Results: 1248 women (51.2 ± 11.2 years) were included: 956 (77.3 %) underwent ABUS+mammography; 283 (29.3 %) ABUS only. Mean follow-up ± SD was 53.5 ± 17.8 month. Thirty-three malignancies were present in the investigated exams. In 28/ 33 cases (84.8 %), lesions were classified BI-RADS 4 or 5 and one (3.6 %) lesion was only visible in ABUS. 3/33 malignancies (9 %) were classified BI-RADS 3. 2/33 (6 %) were visible in mammography and ABUS but not recognized and classified BI-RADS 2 (FN rate 6.1 %). Retrospectively, both cases had "retraction phenomenon sign" in the coronal images. BI-RADS 3 and BI-RADS 4 without a malignancy were attributed to 172 (13.8 %) and 14 (1.1 %) cases, respectively corresponding to a FP rate of 15.3 %. The number of FP as well as the number of addHHUS significantly reduced over the three years (both p < 0.001). Conclusions: After the implementation of an ABUS FP cases and addHHUS reduce over the time.