Related Experiment Videos
Wavelet compression and segmentation of digital mammograms
B J Lucier1, M Kallergi, W Qian
1Department of Mathematics, Purdue University, W. Lafayette, IN.
Journal of Digital Imaging
|February 1, 1994
Summary
Haar wavelets can compress mammographic images effectively, preserving microcalcification details for detection. While high compression introduces artifacts, they are distinguishable from actual calcifications, ensuring diagnostic integrity.
Area of Science:
- Medical Imaging
- Digital Signal Processing
- Radiology
Background:
- Mammography is crucial for early breast cancer detection.
- Image compression is vital for efficient storage and transmission of mammograms.
- Wavelet transforms offer potential for high-quality image compression.
Purpose of the Study:
- To evaluate Haar wavelets for mammographic image compression.
- To assess the impact of compression on microcalcification visualization and detection.
- To determine if compression preserves diagnostic information.
Main Methods:
- Compression of 15 mammograms (105 microns/pixel, 10-12 bits) using Haar wavelets at two rates.
- Expert mammographer evaluation of image quality and diagnostic content.
- Wavelet-based calcification extraction and segmentation on compressed images.
Main Results:
- Good average visualization of microcalcification clusters, degrading with higher compression.
- Compression artifacts were artificial and not mistaken for calcifications.
- Parenchymal density classification remained stable; calcification morphology distorted at high compression.
- Wavelet compression preserved features for successful segmentation of true microcalcification clusters.
- High compression rates achieved without losing critical microcalcification details.
Conclusions:
- Haar wavelets show promise for high-rate mammographic image compression.
- Compression preserves essential features for microcalcification detection.
- Further research in multiresolution analysis for digital mammography is warranted.