Liane Elizabeth Philpotts MD, Presenter: Nothing to Disclose

Sarah Lyla Steenbergen MD, Abstract Co-Author: Nothing to Disclose

Jaime Lynn Geisel MD, Abstract Co-Author: Consultant, Siemens AG

Liva Andrejeva-Wright MD, Abstract Co-Author: Nothing to Disclose

Madhavi Raghu MD, Abstract Co-Author: Nothing to Disclose

Melissa Angeline Durand MD, Abstract Co-Author: Nothing to Disclose

Paul H. Levesque MD, Abstract Co-Author: Nothing to Disclose

Tomosynthesis is a more accurate mammogram resulting in fewer false positives. The purpose of this study was to review recalls from screening tomosynthesis exams to determine which findings represent true lesions and which remain false positives to further refine what constitutes a necessary tomosynthesis recall.

A retrospective review of the breast imaging database was performed to identify all BIRADS 0 (n=295) from 2936 screening (combination 2D and tomosynthesis) exams during a 6 month period (7/1/13 – 12/31/13); recall rate 8.4%. All non-calcification cases (n=157) were retrospectively reviewed on a dedicated workstation by 2 radiologists, with >2 years tomosynthesis experience and the lowest recall rates, who were blinded to further diagnostic imaging and biopsy. The ‘lesions’ in question were assessed for mammographic finding (mass, asymmetry, architectural distortion), seen on 2D (CC, MLO, or both), seen on tomo (CC, MLO, or both), assessment as a true lesion or not, and forced BIRADS and probability of malignancy (1-100%). The outcome after diagnostic work up, and biopsy (if relevant), determined if cases proving to be false positives (BR 1 or 2) could have been prospectively identified.

The forced BIRADS was 61% BR1/2, 24% BR3, 15% BR4/5. After diagnostic workup, the final BIRADS assessment was: 70 (45%)BR1/2, 67 (42%)BR3, 20 (13%) BR4/5. Biopsy was performed in 20 cases revealing 8 cancers. All 8 malignancies presented as masses (7) or architectural distortion (1) and were correctly identified and scored as BIRADS 4b, 4c or 5 on the forced assessment. 36% of cases were felt to be not real on forced assessment (79 % focal asymmetries, 21 % architectural distortion and 83% seen in neither or only 1 view), all of which were benign on final, most (91%) showing normal tissue.

Careful assessment of tomosynthesis screening images permits a high level of confidence in both identifying malignancies and judging normal tissue. Many tomosynthesis recalls could potentially have been avoided.

Even with experience, false positives still occur from tomosynthesis screening as radiologists may have varying levels of confidence in reviewing tomosynthesis images. This study suggests that false positive recalls on screening tomosynthesis can be reduced, in turn reducing unnecessary radiation and cost as well as patient time.

Philpotts, L, Steenbergen, S, Geisel, J, Andrejeva-Wright, L, Raghu, M, Durand, M, Levesque, P, Getting Worked Up Over Nothing: Strategies to Further Reduce False Positives from Tomosynthesis Screening Mammography.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14045665.html