Sample Size and Estimation of Standard Radiation Doses for Pediatric Brain CT

Abstract

Estimation of the standard radiation dose at each imaging facility is required for radiation dose management, including establishment and utilization of the diagnostic reference levels. We investigated methods to estimate the standard dose for pediatric brain computed tomography (CT) using a small number of data. From 980 pediatric brain CT examinations, 25, 50, and 100 examinations were randomly extracted to create small, medium, and large datasets, respectively. The standard dose was estimated by applying grouping and curve-fitting methods for 20 datasets of each sample size. For the grouping method, data were divided into groups according to age or body weight, and the standard dose was defined as a median value in each group. For the curve-fitting methods, logarithmic, power, and bilinear functions were fitted to plots of radiation dose against age or weight, and the standard dose was calculated at the designated age or weight using the derived equation. When the sample size was smaller, the random variations of the estimated standard dose were larger. Better estimation of the standard dose was achieved with the curve-fitting methods than with the grouping method. Power fitting appeared to be more effective than logarithmic and bilinear fittings for suppressing random variation. Determination of the standard dose for pediatric brain CT by the curve-fitting method is recommended to improve radiation dose optimization at facilities performing the imaging procedure infrequently.