Related Experiment Videos
Effect of lossy data compression on quantitative coronary measurements
1Department of Radiology, Leiden University, Medical Centre, The Netherlands.
International Journal of Cardiac Imaging
|August 1, 1997
Summary
Higher lossy data compression ratios negatively impact quantitative coronary arteriography (QCA) results. Using JPEG or LOT compression schemes at ratios 5, 8, or 12 increases measurement variability and is not recommended for QCA.
Area of Science:
- Medical Imaging
- Cardiovascular Interventions
- Image Processing
Background:
- Low compression factors (2-4) are accepted for lossy data compression in Philips DCI.
- The impact of higher lossy compression ratios on quantitative coronary arteriography (QCA) accuracy remains unclear.
Purpose of the Study:
- To assess the influence of LOT and JPEG compression schemes at factors 5, 8, and 12 on coronary measurements.
- To determine if higher lossy compression ratios affect QCA results statistically.
Main Methods:
- 30 original digital coronary artery images were compressed/decompressed using LOT and JPEG at factors 5, 8, 12.
- Automated Coronary Analysis (ACA) package processed original and compressed images.
- 37 obstructed coronary segments were analyzed twice to evaluate intra-observer variability in diameter stenosis.
Main Results:
- Intra-observer systematic errors in diameter measurements were small (<0.07 mm) and not significant across all compression levels.
- Intra-observer random errors increased with compression (up to 0.21 mm) compared to original images (<0.13 mm).
- Statistically significant increases in random errors were observed for JPEG at ratio 5 and LOT at ratio 12.
Conclusions:
- Higher lossy compression ratios (5, 8, 12) negatively affect QCA results due to increased intra-observer variability.
- Inter-compression variability in random errors was high (0.23-0.31 mm).
- Use of JPEG or LOT compression schemes at ratios 5, 8, and 12 is discouraged for QCA.