Improving mass measurement of coronary artery calcification using threshold correction and thin collimation in multi-detector row computed tomography: In vitro experiment

Abstract

Rationale and Objectives. This in vitro study was designed to improve the accuracy of coronary calcium mass measurement from computed tomography (CT) images by developing threshold-based calcium CT number correction and thin-slice spiral techniques. Materials and Methods. A cardiac CT phantom containing simulated calcified cylinders of known calcium density was scanned with sequential 4 × 2.5-, spiral 16 × 1.5-, and spiral 16 × 0.75-mm collimation on 4- and 16-detector row CT scanners. The images obtained from the spiral scans were reconstructed in various slice widths. The calcified cylinders were imaged and their mean CT number and size were measured at thresholds ranging from 50 to 390 Hounsfield unit. The calcium mass measured was compared with the actual value to determine errors, and threshold-based correction factors were derived to minimize the errors. Results. The minimum amount of measurable calcium in 1-mm cylinder was 0.3 mg at the 16 × 1.5-mm protocol and 0.2 mg at the 16 × 0.75-mm protocol. Compared with 2.5-mm sequential protocol, thin-slice spiral protocols yielded a higher radiation exposure and lower or similar image noise levels. The error in calcium mass after correction was significantly smaller than that in measured mass (P < .0001) and was consistent between the imaging protocols (P = .49). The accuracy of mass measurements was clearly improved by using thin-slice imaging protocols especially in 200-mg/cm3 calcium density (P < .0001). Conclusion. The accuracy of calcium mass CT measurement can be improved by threshold-based calcium CT number correction and thin collimation spiral techniques. © AUR, 2003.