Optimization of common CT protocols evaluating statistical measures on real patient images

Abstract

The present work aimed to assess a methodology, in order to propose the minimal exposure levels that yield images suited to diagnostic purposes, for routine computed tomography (CT) studies in three hospitals in Havana, through the evaluation of image quality and exposure levels. The main novelty of this work is that quantitative measures of image quality were calculated over regions of interest on images of real CT studies, with the purpose of determining the influence of dose on image quality. The volume CT air kerma index (CVOL) was measured using the IAEA protocols, and the assessment of clinical criteria proposed in the guidelines was performed by experienced radiologists, which evidenced high potential for optimization. Coherence in the behaviors of statistical metrics determined on real patient images was proven, as well as their ability to compare different scanners in terms of image contrast and noise levels. Finally, the existence of a positive correlation between the statistical image quality measures and volume CT air kerma index values made it possible to determine optimal values of the latter quantity. These results are part of a larger project intended to establish Cuban diagnostic reference levels for CT, and the methodology proved to be readily implementable, as well as applicable to individual institutions.