David Gur ScD, Presenter: Nothing to Disclose

Margarita Louise Zuley MD, Abstract Co-Author: Research grant, Hologic, Inc

Amy H Lu MD, Abstract Co-Author: Nothing to Disclose

Maria Lena Anello MD, Abstract Co-Author: Nothing to Disclose

Grace Y Rathfon MD, Abstract Co-Author: Nothing to Disclose

Gordon Scott Abrams MD, Abstract Co-Author: Nothing to Disclose

Christiane Mary Hakim MD, Abstract Co-Author: Nothing to Disclose

Luisa Paula Wallace MHS, MD, Abstract Co-Author: Nothing to Disclose

Denise Chough MD, Abstract Co-Author: Nothing to Disclose

Marie Adele Ganott MD, Abstract Co-Author: Nothing to Disclose

Ronald L. Perrin MD, Abstract Co-Author: Nothing to Disclose

Jules Henry Sumkin DO, Abstract Co-Author: Scientific Advisory Board, Hologic, Inc

To evaluate screening mammography performance levels using synthetically reconstructed two-dimensional (2D) images in combination with digital breast tomosynthesis (DBT) versus full field digital mammography (FFDM) combined with DBT.

Nine experienced breast imaging radiologists who were specifically trained in reading tomosynthesis retrospectively interpreted an enriched set of 114 mammograms in a fully crossed, balanced study using a screening Breast Imaging-Reporting and Data System (BI-RADS) rating scale when viewing 4 view combination mammography studies under two modes. One mode was the original directly acquired FFDM images combined with DBT and the other mode was the synthetically reconstructed 2D projection images combined with DBT. The reconstructed projection images do not require any additional radiation exposure. We compared the two modes with respect to “sensitivity”, namely a breast with either a pathologically proven cancer (48) or a high risk lesion (6) being recalled; and “specificity”, namely a breast with no abnormality (144) or only benign abnormalities (30) not recalled.

The average sensitivity for actual FFDM with DBT was 0.817 versus 0.770 for synthetic FFDM with DBT, respectively. The use of synthetic images resulted in a decrease in detection of 0.047 with 95% confidence intervals of (-0.002, 0.104) and (-0.01, 0.107) for the study readers or fixed effects (p=0.058), and the general population of readers or random effects, respectively. Recall rates in breasts with no or only benign abnormalities were virtually the same: 0.291 and 0.287 for the two modalities, respectively. Overall performance differences were not statistically significant (p>0.05), but a trend was clearly observed. There were a total of 9 additional micro-calcification related abnormalities (“positive” breast) missed by the readers when interpreting the mode with synthesized images.

Moderately higher sensitivity with the same specificity was achieved when interpreting actual FFDM images as compared with the current version of synthetically generated 2D projection images, both combined with DBT.

Minor improvement in the quality of synthesized 2D images could lead to acceptable diagnostic quality of DBT exams alone; eliminating the need for double exposure during DBT based screening.

Gur, D, Zuley, M, Lu, A, Anello, M, Rathfon, G, Abrams, G, Hakim, C, Wallace, L, Chough, D, Ganott, M, Perrin, R, Sumkin, J, A Two-fold Dose Reduction in Breast Screening with Tomosynthesis (DBT) through the Use of Reconstructed 2D Images from the DBT Dataset?.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11003993.html