An Overcomplete Enhancement of Digital Mammograms.
Abstract
In this report, we describe recent progress in the development of methodology for providing radiologists with a better viewing environment to facilitate the interpretation of mammograms. Our research has been directed towards local processing of images for enhancement of features of diagnostic importance in mammography, and towards development of image fusion methods for blending of enhanced features for an improved visualization of mammograms. We report on the construction of a redundant steerable wavelet transform for multiscale analysis along distinct orientations. We have addressed both elimination of undesirable artifacts commonly associated with wavelet transforms, and efficient implementation of the transform. From qualitative and quantitative comparisons with competitive transforms for fusion of features in mammography, we conclude that the transform is well suited for enhancement of mammograms. By improving the visualization of breast pathology, we can improve chances of early detection while requiring less time to evaluate mammograms for most patients.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1997
- Accession Number
- ADA334425
Entities
People
- Fred Taylor
- Iztok Koren
Organizations
- University of Florida