CT image quality assessment by a Channelized Hotelling Observer (CHO): Application to protocol optimization

In this work we investigate the feasibility of employing a Channelized Hotelling model Observer (CHO) in a CT protocol optimization program with the aim at assuring that the scanners are working at their own best with regard to the quality of images and patient exposure. Although the benefit of using model observers in the clinical protocol optimization is evident, in the practice it is still to be investigated what are the pitfalls associated with this method. With this concern we focused on a clinical protocol for oncology of the abdomen. For the implementation of CHO, we designed a new phantom with the aim of minimizing the number of acquired images. After tuning the model according to a restricted data set, we applied it to the evaluation of a large data set of images obtained with different reconstruction algorithms and acquired on different scanners. Results were very encouraging about the usefulness of CHO for the mentioned purposes. For the first time, at our knowledge, the applicability of CHO was demonstrated for images reconstructed with both filtered back projection (FBP) and iterative (IR) algorithms on the same scanner as well as for images from different scanners, though produced by the same manufacturer. Instead it turned out that CHO was not applicable for the same purposes over images from another manufacturer.