Compression and Classification of Digital Mammograms for Storage, Transmission, and Computer Aided Screening.
Abstract
The general goal of this project is to extend and validate techniques for image compression that we have developed and applied to computerized tomography and magnetic resonance images to the compression of digitized mammograms for efficient transmission, storage, and digital signal processing. This processing will include automatic classification of abnormal tissue so as to permit highlighting of suspicious areas on decompression. A specific goal is to provide compression from 12 bits per pixel (bpp) digital originals to less than 1 bpp while maintaining diagnostic accuracy and clinical utility at least equal to analog film. The performance of compression and classification and the clinical utility of the resulting images will be validated using recently developed clinical simulation and statistical analysis methods. In particular, the diagnostic accuracy of the digitized films at various levels of compression will be compared to that of the analog originals using clinical simulation and statistical analysis methods developed by us.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 21, 1995
- Accession Number
- ADA300012
Entities
People
- Robert M. Gray
Organizations
- Stanford University