Downloaded from http://journals.lww.com/investigativeradiology by BhDMf5ePHKbH4TTImqenVAC0cwDRZApUmmJAvOyfDyS6pkFU3mMld13rt0skVXCo on 01/26/2019 Synthetic 2-Dimensional Mammography Can Replace Digital Mammography as an Adjunct to Wide-Angle Digital Breast Tomosynthesis Paola Clauser, MD,* Pascal A.T. Baltzer, MD,* Panagiotis Kapetas, MD,* Ramona Woitek, MD,* Michael Weber, PhD,† Federica Leone, MD,‡ Maria Bernathova, MD,* and Thomas H. Helbich, MD* Objectives: The aim of this study was to evaluate the detection rate and diagnos- tic performance of 2-dimensional synthetic mammography (SM) as an adjunct to wide-angle digital breast tomosynthesis (WA-DBT) compared with digital mam- mography (DM) alone or to DM in combination with WA-DBT. Materials and Methods: There were 205 women with 179 lesions included in this retrospective reader study. Patients underwent bilateral, 2-view (2v) DM and WA-DBT between March and June 2015. The standard of reference was his- tology and/or 1-year stability at follow-up. Four blinded readers randomly evalu- ated images according to the BI-RADS lexicon from 3 different protocols: 2v DM alone, 2v DM with 2v WA-DBT, and 2v SM with 2v WA-DBT. Detection rate, sensitivity, specificity, and accuracy were calculated and compared using multivariate analysis. Readers' confidence and image quality were evaluated. Results: The detection rate ranged from 68.7% to 79.9% for DM, 76.5% to 84.4% for DM with WA-DBT, and 73.2% to 84.9% for SM with WA-DBT. Sensitivity and accuracy were significantly higher when DBT was available (P < 0.001). Specificity did not differ significantly between DM only, DM with WA-DBT, or SM with WA-DBT (P ≥ 0.846). Wide-angle DBT combined readings did not differ between SM and DM in terms of sensitivity, specificity, and accuracy (P ≥ 0.341). Readers' confidence and image quality was rated good to excellent. Conclusions: Wide-angle DBT combined with DM or SM increases sensitivity and accuracy without reducing specificity compared with DM alone. Wide-angle DBT combined readings did not differ between SM and DM; therefore, SM should replace DM for combined readings with WA-DBT. Key Words: breast neoplasms, digital breast tomosynthesis, digital mammography, synthesized mammography, sensitivity, specificity (Invest Radiol 2019;54: 83–88) D igital breast tomosynthesis (DBT) in combination with digital mammography (DM) has been shown to significantly increase cancer detection rate. 1–10 Reconstructed, synthesized 2-dimensional (2D) mammograms (synthetic mammography [SM]) from the DBT dataset have been devel- oped 11 to decrease radiation exposure by up to 45%, while maintaining the diagnostic benefits of 2D mammography and DBT. 12–16 The added value of SM has been evaluated exclusively using narrow-angle DBT (NA-DBT) systems. 12–14,17–19 Meanwhile, wide-angle DBT (WA- DBT) systems, which allow reconstruction of synthesized 2D mammo- grams as well, have been introduced. 20,21 To the best of our knowledge, no data about the detection rate and diagnostic performance of SM obtained from a WA-DBT system have been reported as yet. Thus, the purpose of our study was to assess the detection rate and the diagnostic performance of SM with WA-DBT compared with DM alone and DM in combination with WA-DBT. MATERIALS AND METHODS The study was supported by a grant from Siemens Healthcare (Erlangen, Germany). All authors had full control of all data and statistical results. Eligible subjects for this institutional review board–approved, retrospective study were patients who had undergone bilateral, 2-view (2v) DM and 2v WA-DBT as a screening or diagnostic examination at our institution between March and June 2015. Overall, 285 patients were examined. Due to the retrospective nature of the study, the need for an informed consent was waived. The exclusion criterion was the absence of a standard of refer- ence (image-guided biopsy, surgery, or at least 12 months of follow- up; Fig. 1). Included in the analysis were 205 women (mean age, 56.2 years; range, 36–84): 49 presented with no mammographic findings (BI- RADS 1), whereas 156 had at least 1 lesion (BI-RADS 2–5). Nineteen patients (9.3%) had, overall, 2 lesions, and 2 patients (1%) had 3 lesions. Finally, 179 lesions were included in the analysis, 89 malignant and 90 benign. Lesion histology and type of lesions included are shown in Table 1. The ACR BI-RADS 22 breast density distribution was as follows: ACR-a in 31 patients (15.1%), ACR-b in 86 (42.0%), ACR-c in 70 (34.1%), and ACR-d in 18 (8.8%). Image Acquisition Digital mammography and WA-DBT were acquired in 2 views (craniocaudal and mediolateral oblique) with a commercially available system (Mammomat Inspiration; Siemens Healthcare GmbH, Erlangen, Germany) during the same breast compression. This device is character- ized by a W/Rh anode/filter combination. The WA-DBT view acquires 25 projections over an angular range of 50 degrees and the automatic ex- posure control was set to deliver a dose twice that of DM. The WA-DBT projections were reconstructed using EMPIRE technology (Siemens Healthcare GmbH, Erlangen, Germany), including statistical artifact re- duction to mitigate out-of-plane artifacts and iterative filtering in image space to suppress noise. 23 The resulting tomosynthesis slices have an in-plane resolution of 0.085 mm 0.085 mm and are 1 mm apart. From the WA-DBT volumes, synthesized mammograms (Insight 2D; Siemens Healthcare GmbH, Erlangen, Germany) were reconstructed based on a 3D volume ray-casting method, to obtain the exact same distri- bution of microcalcifications and the same tissue structures as in the digital mammograms. Received for publication July 12, 2018; and accepted for publication, after revision, August 14, 2018. From the *Department of Biomedical Imaging and Image-Guided Therapy, †Division of General and Pediatric Radiology, Department of Biomedical Imaging and Image-Guided Therapy, Medical University of Vienna, Vienna, Austria; and ‡ASST Fatebenefratelli-Sacco, Ospedale Fatebenefrateli e Oftalmico, Milan, Italy. Conflicts of interest and sources of funding: The study was sponsored by Siemens Healthcare (T.H.H.). P.C. and M.B. have received speaker honoraria from Siemens Healthcare. For the remaining authors, none were declared. Correspondence to: Thomas H. Helbich, MD, Department of Biomedical Imaging and Image-Guided Therapy, Medical University of Vienna, General Hospital Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria. E-mail: thomas. helbich@meduniwien.ac.at. Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0020-9996/19/5402–0083 DOI: 10.1097/RLI.0000000000000513 ORIGINAL ARTICLE Investigative Radiology • Volume 54, Number 2, February 2019 www.investigativeradiology.com 83 Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.