False-Negative Interpretations in a CAD Environment
Abstract
The purpose of this project is to examine the impact of CAD schemes on the diagnostic performance of radiologists, especially the change of false-negative interpretations under a CAD cueing environment. Using an observer performance study, we hope to better understand the impact of CAD schemes on diagnostic performance for detection of subtle abnormalities and to find an optimal approach to use CAD schemes in a screening environment. In this project, seven radiologists participated in the study. Each interpreted 120 subtle mammographic cases in five different reading modes (including a non-cued mode and four CAD-cued modes) . The study results demonstrated that CAD cueing with high accuracy (i.e., 90% sensitivity and 0.5 false-positive per image) significantly improved observer's performance (p<0.05) . As accuracy of CAD cueing decreased so did observer performance. Reducing cueing sensitivity and specificity increased false-negative rates in non-cued areas (p<0/05) . In conclusion, this study indicates that in a laboratory environment, observer performance in the detection of subtle mammographic abnormalities is significantly affected by the inherent performance of a CAD cueing system.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2001
- Accession Number
- ADB274470
Entities
People
- Bin Zheng
Organizations
- University of Pittsburgh