Computer Aided Detection and Digital Mammography: CAD Study I and CAD Study II

Principal Investigator: Etta D. Pisano, MD

Status:  In Analysis

Main Objective: CAD I: To estimate the sensitivity of each of the CAD software systems to detect known cancers when applied to digitally acquired mammograms (digital CAD) and to digitized screen film mammograms (analog CAD). CAD II: To assess whether the use of CAD can improve the sensitivity, specificity, and AUC of digital mammograms obtained from the ACRIN DMIST archives when reviewed by several radiologists.

Participants: CAD Study I will use all available known cancer cases from the DMIST population for whom both the screen-film and raw digital images are available and are compatible with the CAD software.  CAD Study II will use a random sample of cases stratified first by cancer status and then by the BI-RADS score assigned to the screen-film mammogram from the initial DMIST screening.

Study Design Summary: CAD I: Computer Aided Detection (CAD) algorithms were applied to the digital mammograms of the DMIST cases known to have breast cancer, and to the corresponding screen- film mammograms. Softcopy digital mammograms were loaded onto various manufacturers' CAD systems for CAD processing. One radiologist reviewed the CAD-processed images, and recorded whether the system identified the known cancer as an area of concern. In addition, the radiologist recorded the number of false positive marks CAD placed on each view. CAD II: A controlled reader study will be conducted with 45 radiologist readers, 15 for each CAD vendor.  Each reader will read up to 300 cases in total, first without and then with CAD applied, all on softcopy digital display.