A method of predicting the image noise in paediatric multi-slice computed tomography images

Abstract

The purpose of this study was to develop an equation with which to determine the tube current to be used in order to obtain a certain image noise level for differently sized children undergoing multi-slice computed tomography examination. The relationship between image noise and detector dose for different examination protocols was established for a LightSpeed Ultra, an eight slice CT from GEMS, using homogeneous water phantoms of different sizes. Three different anatomical areas (head, thorax and abdomen) were studied in 111 patients between 0 and 17 y of age. The mean ratio between the calculated and the measured noise in patient images was established for the different areas. Head examinations showed the best correlation (measured-to-calculated noise ratio = 1.01). In the thorax, the calculated noise was generally higher than the measured noise (ratio = 0.74), and in the abdomen, the opposite result was found (ratio = 1.20).

Pediatric Dose Reduction in Computed Tomography

“…4), but could result in false positives as well. Work in trying to determine tube current reduction (e.g., noise) thresholds for pediatric CT using investigational noise addition techniques is currently attempting to address these thresholds (Frush et al 2002;Fefferman et al 2005;Ledenius et al 2005). These noise addition technologies obviate multiple scans in children at different tube currents since the manipulation is of MDCT scan data.…”

Nonlinear Model of Image Noise: An Application on Computed Tomography including Beam Hardening and Image Processing Algorithms

“…The contribution of F 2 to noise is the same for both 120 kV and 135 kV. Bearing in mind the confidence intervals of the nonlinear parameters (see Table 3), the proposed model does not present significant differences with the parameters of quantitative factors with regard to other reported models [1]- [3], where the noise is directly proportional to the inverse of the square root of the mAs and the pixel size. The parameter of 4 F does not include the value −0.5 as it could be expected [2].…”

“…Other important influences, which lead to image quality variations, such as beam hardening and reconstruction algorithms implemented by manufacturers can no longer be described by such a simple equation. A correlation between image noise and dose for each examination protocol was determined by measuring the noise while keeping the scanning parameters other than tube current constant in [3] [4]. A more inclusive noise model could provide knowledge about quantitative effect of reconstruction algorithms and beam hardening and a predictive tool for clinical applications.…”