ABSTRACT
This chapter provides background material, starting with diagnostic interpretations occurring every day in hospitals, which define diagnostic tasks. The process of imaging device development by manufacturers is described, stressing the role of physical measurements in optimizing the design. Device approval for marketing requires measurements falling under the rubric of observer performance studies, the subject of this book. Once the device is deployed, hospital-based medical physicists use phantom quality control measurements to maintain image quality. Lacking the complexity of clinical images, phantom measurements may not detect clinical image quality degradation. Model observers, that reduce the imaging process to mathematical models and formulae, are intended to bridge the gap. Unlike physical, phantom and model observer measurements, observer performance methods measure the effect of the entire imaging chain, including the critical role of the radiologist. Four observer-performance paradigms are described that account for localization information to different extents. Physical and observer performance methods are put in the context of a hierarchy of efficacy levels, where the measurements become increasingly difficult, but more clinically meaningful, as one moves to higher levels. An overview of the book, which is split into four parts, is presented and suggestions are made on how to best use it. An online appendix introduces the reader to R/RStudio, which are used as pedagogical tools throughout the book, and RJafroc, an R-package encapsulating all analytical methods described in this book. It was coded mainly by Xuetong Zhai, MS. It extends the capabilities of the author’s earlier Windows software (JAFROC at https://www.devchakraborty.com">https://www.devchakraborty.com). All online material now resides at https://expertcadanalytics.com">https://expertcadanalytics.com.
