Detection of masses and clustered microcalcifications on data compressed mammograms: an observer performance study

Area of Science:

• Radiology and Medical Imaging
• Digital Mammography
• Image Compression

Background:

• Mammography is crucial for early breast cancer detection.
• Digital mammography allows for image compression to reduce file size and transmission time.
• The impact of varying data compression levels on diagnostic accuracy requires evaluation.

Purpose of the Study:

• To assess how data compression affects radiologists' ability to detect breast masses and clustered microcalcifications.
• To quantify the impact of different compression ratios on observer performance in mammogram interpretation.

Main Methods:

• An observer performance study involving eight radiologists interpreting 60 mammographic images.
• Images were presented in six modes, from uncompressed to a 101:1 compression ratio.
• Radiologists rated the likelihood of masses and microcalcifications and subjective image quality; receiver operating characteristic (ROC) analysis was performed.

Main Results:

• A significant decrease in diagnostic performance for clustered microcalcifications was observed at the highest data compression level (p < 0.01).
• A statistically significant trend of decreasing performance with increasing compression was noted for microcalcification detection (p < 0.05).
• No significant differences in mass detection performance were found across all tested data compression levels.

Conclusions:

• High levels of mammogram data compression degrade observer performance in detecting clustered microcalcifications.
• Data compression methods and levels evaluated in this study did not impact the detection of breast masses.