Antonius A.J. Roelofs PhD, Abstract Co-Author: Nothing to Disclose

Nora Wedekind MS, Abstract Co-Author: Nothing to Disclose

Sander van Woudenberg, Abstract Co-Author: Nothing to Disclose

Christian Beck, Abstract Co-Author: Nothing to Disclose

J.H.C.L. Hendriks PhD, Abstract Co-Author: Nothing to Disclose

Nico Karssemeijer PhD, Abstract Co-Author: Nothing to Disclose

C.J.G. Evertsz PhD, Presenter: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

To investigate the impact of prior mammograms on mammography screening performance in a soft-copy reading environment.

Ten radiologists experienced in screening mammography each read 160 digitized cases twice, once with and once without prior-round screening mammograms, using a dedicated mammographic review station (MeVis BreastCare). Fifty percent of the cases were screening mammograms of cases in which breast cancer was detected at a later stage; the others were normals and cases with benign abnormalities. Only positive cases were selected in which abnormalies were mammographically visible in retrospect. Readers marked locations of potential abnormalities, characterized them according to BI-RADS main categories, and estimated likelihood of malignancy of each lesion. They also classified the prior mammograms as "needed", "helpful" or "not needed". During reading CAD markers (R2 Technology) for microcalcifications were accessible. Cases were presented in a balanced order regarding the availability of prior mammograms. Between the first and second reading of a case was a minimal period of two weeks. All readers participated in soft-copy reading training before starting the study. Additionally, they read 20 cases to get familiar with the use of the malignancy rating scale. Effects of prior mammograms on detection performance were determined using MRMC-ROC and LROC analysis, calculation of sensitivity/specificity, and by analyzing radiologists' opinions regarding the necessity of the priors. Results were also analyzed separately for the main BI-RADS characterization classes.

Reading performance was significantly better when the prior screening mammogram was available. LROC analysis showed that the number of false positives was roughly two times less with the use of priors at a fixed value of sensitivity. Radiologists had difficulty in selecting those cases in which prior mammograms were beneficial.

Prior mammograms have a large impact on detection performance in screening mammography. Radiologists find it hard to select a subset of cases in which they should take priors into account. This suggests that prior mammograms should always be viewed.

C.B.: in an employee of MeVis BreastCare.N.K.: is a shareholder, consultant, and has received grant support from R2 Technology.N.W.: is an employee of MeVis.C.E.: is on the board of MeVis BreastCare; this presentation received a European Commission grant for the screen trial project, in which MeVis BreastCare participates.

Roelofs, A, Wedekind, N, van Woudenberg, S, Beck, C, Hendriks, J, Karssemeijer, N, Evertsz, C, et al, , The Impact of Prior Mammograms in Mammography Screening.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4403992.html