Computer-Aided Diagnosis and Feature-Guided Data Reduction Systems in Mammography.

摘要

We have been developing CAD algorithms in detection of microcalcifications and masses using advanced image processing and computer vision techniques. Our CAD algorithms have provided very promising results in laboratory tests. The goals in this project are to implement our CAD algorithms in a fast workstation, develop user interfaces for efficient operation of the CAD programs, and conduct a pilot clinical trial of the CAD schemes at two mammographic screening sites. Based on the results of the pilot clinical trial, we can evaluate the sensitivity and specificity of the CAD algorithms, analyze the effects of the CAD schemes on manmographic screening, identity potential problems in a clinical environment, and develop methods to further improve the CAD schemes in the future. We believe that this is a crucial step to develop a clinically practical CAD workstation.