Investigation of Three-Group Classifiers to Fully Automate Detection and Classification of Breast Lesions in an Intelligent CAD Mammography Workstation
Abstract
Our goal is to develop a fully automated classification scheme for computer-aided diagnosis in mammography. Our proposed scheme would classify computer detections into three groups: malignant lesions, benign lesions, and false-positive computer detections. This method presents considerable difficulties in terms of both signal detection theory and performance evaluation methods. During the past year we have focused our efforts on theoretical understanding of the behavior and properties of the three-group classifier. We have proven that an obvious generalization of the well-known two-group performance metric is not in fact a useful performance metric for classification tasks with three or more groups. We have developed an evaluation technique, independent of those we have previously developed, for assessing the accuracy of estimates of ideal observer decision variables. We have analyzed several recently proposed three-group classification methods in terms of the three-group ideal observer. Finally. we have shown that the three decision boundary lines used by the three-group ideal observer are not arbitrary, but are intricately related to one another. A three-group classifier could potentially allow radiologists to detect more malignant breast lesions without increasing their false-positive biopsy rates.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2005
- Accession Number
- ADA437311
Entities
People
- Charles E. Metz
- Darrin C. Edwards
- Maryellen Lissak Giger
Organizations
- University of Chicago