Per Skaane MD, PhD, Presenter: Grant, Hologic, Inc

Randi Gullien, Abstract Co-Author: Nothing to Disclose

Ellen B. Eben MD, Abstract Co-Author: Nothing to Disclose

Ulrika Ekseth MD, Abstract Co-Author: Nothing to Disclose

Unni Haakenaasen MD, Abstract Co-Author: Nothing to Disclose

Gunnar Jahr MD, Abstract Co-Author: Nothing to Disclose

Ingvild Naess Jebsen MD, Abstract Co-Author: Nothing to Disclose

Mona Krager MD, Abstract Co-Author: Nothing to Disclose

To compare the interpretation times for full-field digital mammography (FFDM) and digital breast tomosynthesis (DBT) in an organized population-based mammography screening program with independent double reading.

The study was approved by the Ethical Committee, and all women signed a written consent. Women 50-69 years of age attending the screening program were offered 2-view DBT in addition to the standard 2-view FFDM. There was independent double reading in the FFDM arm, one reader having access to CAD and independent double reading in the DBT arm, one reader having access to a synthetic FFDM (created from the DBT) instead of conventional FFDM. Thus, each exam was interpreted independently by 4 readers. Interpretation included a 5-point rating scale for probability of cancer. Interpretation time was recorded for all exams, starting when the case report form was presented and closed when clicking the mouse for conclusion. All positive scores were discussed in a consensus meeting before recall decision.

During the first 3 months of the trial, 4,534 women underwent FFDM, of which 3,356 (74%) were examined with DBT. Excluding the reading times for the first month (learning curve), readers with few interpretations, and outliers, the mean interpretation time (all readers) for FFDM (excluding arm with CAD) was 49 sec. (range, 32-81) and the mean reading time for DBT (excluding synthetic 2D arm) 92 sec. (range, 37-157). Of the 33 cancers detected so far, 25 cancers had independent double reading for both FFDM and DBT. 8 cancers were missed in the FFDM arm and none in the DBT arm. Thus, the relative increase in cancer detection using DBT was 47% (17+8/17). There were 9 individual false negative interpretations in the DBT arm, of which 6 had shorter interpretation time than the mean of 92 sec.

Reading time for DBT (mean 86 sec.) is longer than for FFDM but would be acceptable even in a high-volume screening program. The shorter interpretation time for most of the false negative DBT interpretations and the considerable interobserver variation indicates a learning curve for DBT reading in a batch mode. 

DBT is a promising technique in breast cancer screening. Interpretation time for DBT is acceptable for high-volume batch mode screening.

Skaane, P, Gullien, R, Eben, E, Ekseth, U, Haakenaasen, U, Jahr, G, Jebsen, I, Krager, M, Reading Time of FFDM and Tomosynthesis in a Population-based Screening Program.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11011027.html