Current Status of Computerized Decision Support Systems in Mammography

  • G.D. Tourassi
Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 184)


Breast cancer is one of the most devastating and deadly diseases for women today. Despite advances in cancer treatment, early mammographic detection remains the first line of defense in the battle against breast cancer. Patients with early-detected malignancies have a significantly lower mortality rate. Nevertheless, it is reported that up to 30% of breast lesions go undetected in screening mammograms and up to 2/3 of those lesions are visible in retrospect. The clinical significance of early diagnosis and the difficulty of the diagnostic task have generated a tremendous interest on developing computerized decision support systems in mammography. Their main goal is to offer radiologists a reliable and fast “second opinion”. Several systems have been developed over the past decade and some have successfully entered the clinical arena. Although several studies have indicated a positive impact on early breast cancer detection, the results are mixed and the decision support systems are under ongoing development and evaluation. In addition, there are still several unresolved issues such as their true impact on breast cancer mortality, the overall impact on the recall rate of mammograms and thus the radiologists’ workload, the reproducibility of the computerized second opinions, the ability of a knowledgeable radiologist to effectively process these opinions, and ultimately clinical acceptance. Furthermore, the medical and legal implications of storing and/or dismissing computerized second opinions are currently unknown. Given the number of unresolved issues, the clinical role of the decision support systems in mammography continues to evolve. The purpose of this article is to review the present state of computer-assisted detection (CAD) and diagnosis (CADx) in mammography. Specifically, the article will describe the principles of CAD/CADx, how it is currently applied in mammography, examine reported limitations, and identify future research directions. Research work is presented towards the application of knowledge-based systems in mammography to address some of the current CAD limitations. Finally, the natural extension of CAD to telemammography is discussed.


Decision Support System Receiver Operating Characteristic Digital Mammography Architectural Distortion Screen Mammogram 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Authors and Affiliations

  • G.D. Tourassi
    • 1
  1. 1.Department of RadiologyDuke University Medical CenterDurham

Personalised recommendations