Computer-Aided Classification of Malignant and Benign Lesions on Mammograms
Abstract
The purpose of this project is to develop computerized classification methods for mammographic abnormalities which will aid radiologists in deciding whether a patient should be biopsied. The regions of interest (ROIs) will be identified by radiologists, and the features to be used in classification will be computer-extracted image features. In the second year of our project, we have made progress in all five major objectives of the proposal. We have digitized over 600 new films for computerized analysis. We have investigated new segmentation and morphological feature extraction methods for classification of mammographic masses. We have investigated a new high-sensitivity training algorithm for artificial network, and designed a new hierarchical classifier for improved separation between malignant and benign cases. The mass classification algorithm that was developed in the first year of the project was evaluated with six radiologists in an observer study. The results of the observer study indicate that their classification accuracy was significantly improved when they were aided by our computer classification scores. Finally, using an artificial neural network classifier, texture features were evaluated for classification of 86 ROIs containing microcalcification clusters.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1998
- Accession Number
- ADA350905
Entities
People
- Berkman Sahiner
Organizations
- University of Michigan