JPEG2000 Compression of CT Images Used for Measuring Coronary Artery Calcification Score: Assessment of Optimal Compression Threshold
AJR:198, April 2012
Tae Ki Kim Kil Joong Kim Sang II Choi Kyoung Ho Lee Suyoung Kim Jae Hwa Cho Jong HoarYoon
OBJECTIVE. The purpose of our study was to assess the acceptable compression threshold for JPEG2000 compression of CT images used for measuring coronary artery calcification scores (CACS) in terms of variability.
MATERIALS AND METHODS. In a retrospective review, 80 patients who had undergone CT for determination of the CACS were compiled in four subsets (20 scans each) according to CACS: 0, subset A; > 0 to > 100, subset B; > 100 to < 400, subset C; and > 400, subset D. Each scan was compressed using eight compression ratios (CRs). We measured the CACS on all 720 CT scans (80 original and 640 compressed scans). For each compressed scan, the variability in CACS was evaluated by comparing with the CACS of the corresponding original CT scan.
RESULTS. For each subset and each CR, we determined whether the upper limit of the onesided 95% CI of the variability in CACS exceeded 5%. The variability in CACS tended to increase as the CR increased and tended to decrease in the order of increasing CACSs at each CR (i.e., subset B > subset C > subset D). With 5% as the limit of variability, acceptable compression CRs were between 20:1 and 25:1 for subset B; between 40:1 and 60:1 for subset C; and > 100:1 for subset D.
CONCLUSION. A level of 20:1 could be a potentially acceptable threshold for JPEG2000 compression of CT images used for measuring CACS, with 5% of the variability in CACS as the acceptable limit of variability.