Related Experiment Videos
Comparing the performance of mammographic enhancement algorithms: a preference study.
R Sivaramakrishna1, N A Obuchowski, W A Chilcote
1Department of Biomedical Engineering, The Cleveland Clinic Foundation, OH 44195, USA.
AJR. American Journal of Roentgenology
|July 6, 2000
Summary
Image enhancement algorithms improved microcalcification visibility in mammograms, with adaptive neighborhood contrast enhancement being most preferred. However, no algorithm significantly enhanced mass detection compared to unenhanced images.
Area of Science:
- Radiology and Medical Imaging
- Digital Mammography
- Image Processing
Background:
- Mammography is crucial for breast cancer detection.
- Secondary digitization of film mammograms is common.
- Image enhancement may improve diagnostic accuracy.
Purpose of the Study:
- To compare four image enhancement algorithms for digitized mammograms.
- To evaluate algorithm performance on masses and microcalcifications.
- To assess performance in a clinical soft-copy display setting.
Main Methods:
- Four algorithms (adaptive unsharp masking, contrast-limited adaptive histogram equalization, adaptive neighborhood contrast enhancement, wavelet-based enhancement) were applied.
- Digitized mammograms with known pathology (masses, microcalcifications) were used.
- Expert mammographers ranked enhanced and unenhanced images.
Main Results:
- Adaptive neighborhood contrast enhancement was preferred for microcalcifications (49%).
- Unenhanced images were preferred for masses (58%).
- No significant improvement for masses with enhancement algorithms.
Conclusions:
- Image enhancement can improve microcalcification visibility.
- Adaptive neighborhood contrast enhancement is effective for microcalcifications.
- No tested algorithm significantly improved mass detection over unenhanced images.