Accuracy in quantification of coronary calcification with CT: a cork-dog heart phantom study

Rationale and objectives: Coronary artery calcium is a sensitive risk predictor of cardiac events. However, measurement of calcium foci is affected by partial-volume effects, which ultimately have an effect on accuracy and reproducibility of calcium scores. In this study, we describe the accuracy of quantification of calcium foci of known size and density using cork-dog heart phantoms.

Materials and methods: Five study phantoms were constructed from cork chests and dog hearts containing 135 calcium hydroxyapatite (CaHA) foci of known volume, mass, and concentration located in the coronary arteries or the myocardium. Hearts were separated into two groups: (1) three hearts containing large, high-density foci and (2) two hearts containing small, low-density foci. The phantoms were scanned using a standard coronary artery calcium (CAC) protocol and the volume and mean intensity of foci were measured.

Results: In group 1, the total volume of 87 CaHA foci measured was 4284 and 3779 mm(3) with electron beam computed tomography (EBCT); multidetector computed tomography (MDCT), respectively (P < .001). Both were significantly larger than the true volume (2713.9 mm(3), P < .001). In Group 2, the total volume of 57 CaHA foci measured was 592.6 and 702.9 mm(3) with EBT and MDCT, respectively (P < .001). Both were significantly smaller than the true volume (1733.2 mm(3), P < .001). We found that EBCT values for volume were approximately generally higher than MDCT values, but strongly correlated (r = 0.95, P < .0001). Agatston scores were found to be nearly equivalent between EBCT and MDCT and were similarly strongly correlated (r = 0.97, P < .0001).

Conclusions: Computed tomography images overestimate the volume of large, dense CaHA foci while underestimating the volume of smaller (<6.6 mm(3)), less dense foci. This may have significant implications on CAC scoring and volume measurement. EBCT overestimated calcium more than MDCT, most likely from increased image noise.