Automatic detection and quantification of the Agatston coronary artery calcium score on contrast computed tomography angiography
OBJECTIVE To study how much the calcium scores at various phases throughout the cardiac cycle deviate from the score in the most motionless phase during retrospectively electrocardiogram (ECG)-gated multidetector row computed tomography (MDCT) of the heart and to evaluate how to optimize ECG-based tube current modulation so that errors in calcium scoring can be minimized while dose savings can be maximized. MATERIALS AND METHODS In 73 subjects with known or suspected coronary artery disease we performed retrospectively ECG-gated 64-detector row computed tomography for calcium scoring. Four subjects were excluded after scanning because of breathing artifacts or lack of coronary calcification. The scans of 69 subjects (46 men, mean age 62 +/- 6 years) were used for further analysis. Heart rate during the scan was recorded. In each patient, calcium scoring [Agatston score (AS), mass score (MS), and volume score, (VS)] was performed on 10 data sets reconstructed at 10%-intervals throughout the cardiac cycle. The most motionless phase was subjectively determined and used as the reference phase. For the score in each phase, deviation from the score in the reference phase was determined. An ECG-simulator was used to determine the amount of dose saving while scanning with dose modulation and applying diagnostic dose during 1 or several phases. RESULTS Mean heart rate was 63 (+/-13) beats per minute (bpm). In 51% of patients the reference phase was the 70% phase. Using the calcium score in the 70% phase (mid-diastole) instead of the reference at heart rates below 70 bpm would have induced a median score deviation of 0% [interquartile range: 0%-6% (AS, MS, and VS)] and using the calcium score in the 40% phase (end-systole) at heart rates > or =70 bpm would also have induced a median score deviation of 0% [interquartile range: 0%-7% (AS), 0%-5% (MS), and 0%-3% (VS)]. Errors in calcium scores of more than 10% occur in around 10% of subjects for all 3 scoring algorithms. Dose savings increased with lower heart rates and shorter application of diagnostic dose. CONCLUSIONS The optimum phases for dose modulation are 70% (mid-diastole) at heart rates below 70 bpm and 40% (end-systole) at heart rates above 70 bpm. Under these conditions dose saving is maximum and a median error of 0% is found for the various calcium scoring techniques with score errors of more than 10% in around 10% of subjects.