ABSTRACT
There are three components to objective assessment of image quality: (1) the task, (2) the images, and (3) the observer(s). The task is connected with the type of observer performance method used and can include signal detection, parameter estimation, or combined variants of the two. The images are obtained in a way that enables the investigation of specific tasks. The ability of an observer to perform a relevant task is usually limited by the noise sources in the imaging chain, which includes the anatomical noise present in most images. The image quality concept is thus inherently statistical and based on variations in the image formation and on observer-related variations (inter and intra). As a result, there is a strong relationship between these three components and the subsequent observer performance. The observer can be a human reader, such as a radiologist, or a mathematical model, also called an “observer model.” Mathematical observer models for tomosynthesis are described in Chapter 10. This chapter deals with the objective assessment of medical images with human observers. Such studies are often called “observer studies” or “observer experiments.”
