Anatomical noise model for ct head images: Preliminary results

Abstract

The availability of an anatomical predictive noise model provides new knowledge about the physical factors tradeoff and the quantitative effect of multiple factors over the diagnostic image quality. Here, the anatomical noise was defined as the standard deviation of pixel intensities, contained in a region located at the center of an axial CT pediatric head image. Images were obtained by using an Automatic Exposure Control system. The purpose of this work was to determine the association between the noise on diagnostic images and a noise model based on phantom measurements. The model of anatomical noise obtained has an adequate predictive value, with a correlation coefficient of 0.97 (significance level of 95%) and mean square error of 1.9 × 10-6.