A CT acquisition technique to generate images at various dose levels for prospective dose reduction studies

OBJECTIVE. The purpose of this article is to determine whether the average of N CT images acquired at a particular dose (D) has image noise equivalent to that of a single image acquired at a dose of N x D. MATERIALS AND METHODS. An electron density phantom, an image quality phantom, and an adult anthropomorphic phantom were scanned multiple times on a 16-MDCT scanner at five effective tube current-rotation time product (mAs) settings (130 kVp; 12, 24, 48, 72, and 144 mAs). Lower-mAs images were averaged to simulate higher-mAs images. Differences in CT number and image noise between simulated and acquired images were quantified using the electron density phantom. Image quality phantom images were scored by three physicists to investigate differences in low- and high-contrast resolution. A forced-choice observer study was performed with three radiologists using anthropomorphic phantom images to evaluate differences in overall image quality. RESULTS. The CT number was, on average, reproduced to within 1 HU, and image noise was reproduced to within 4%, which is below the threshold for visibly perceptible differences in noise. Low- and high-contrast resolution were not degraded, and simulated images were visually indistinguishable from acquired images. CONCLUSION. For the dose range studied, it was concluded that the image quality of a CT image produced by averaging multiple low-mAs CT images is identical to that of a high-mAs image acquired at equivalent effective dose, when all other acquisition and reconstruction parameters are held constant. Prospective CT dose-reduction studies may be feasible by acquiring multiple low-dose scans instead of a single high-dose scan. Simulated high-dose images could be interpreted clinically, whereas lower-dose images would be available for an observer study.