The broad long-term objective is to reduce deaths from breast cancer by improving cancer screening through the use of direct digital mammography. The diagnostic accuracy of mammographers with conventional screen-film mammograms (SFM) and high-resolution full breast direct digital film (DFM) and digital video (DVM) mammograms will be compared and quantified in a probability-based sample of women (n=100) with verified presence or absence of breast cancer.
The specific aims (SA) are: (SA1) To compare and test for differences in the diagnostic accuracy of mammographers (n=16) with three methods of screening mammography (SFM, DFM & DVM); (SA2) To determine if mammographer expertise alters performance with digital mammography (DFM &DVM) when compared to conventional screen-film mammography (SFM); and (SA3) To determine if digital mammography will improve screening for breast cancer in women problematic for SFM, specifically those with dense breast tissue or false alarm (false positive) mammograms in general. Additional aims will evaluate individual mammographer diagnostic accuracy with digital mammography, relative performance of digital mammography with important breast characteristics and mammographic signs of cancer (density, calcifications, architectural distortion, and masses) and mammographers' subjective rating of image quality. The health-relatedness is that digital mammography may significantly improve breast cancer screening with mammography, allow computer-aided diagnosis, reduce the cost of screening mammography, and improve screening availability to currently underserved populations. Quantifying the diagnostic performance of mammographers with digital mammograms is a critical step needed prior to implementation of this technology. The research design is a multiple observer, blinded, randomized, comparison of the current standard (SFM) to two methods of digital mammography (DFM & DVM). The study population will be probablistically sampled from a completed prospective study of digital mammography. Results will be analyzed using receiver operating characteristic methodology and linear modeling to describe mammographer performance in addition to basic measures of diagnostic accuracy.
