Image Processing in Computed Radiography

This article starts with an overview of image processing techniques used in storage phosphor based Computed Radiography (CR) systems. Next it elaborates on a selection of image enhancement algorithms. Both the working principles and image quality issues are discussed. The main focus is on multiscale image enhancement, which has become state-of-the-art. Introduction Since the early days of CR technology developers have investigated solutions for bridging the gap between the very large dynamic range that characterizes the CR detector and the limited range of the output medium and viewing process. A considerable part of image processing functionality in current CR systems deals directly or indirectly with the issue of manipulating image contrast, in such a way that all relevant image features are rendered to an appropriate level of visibility, despite the restriction of viewing density range. CR equipment manufacturers have adopted basic image processing techniques, or they have developed dedicated solutions. With this article it is our aim to provide a better understanding of the essential image enhancement techniques of CR systems, what their purpose is, how they operate, and how they affect image quality. CR image processing OVERVIEW A simplified diagram of the image processing operations in current CR systems is depicted in Fig. 3.1. The ensemble of operations applied to the stream of image data could be roughly entitled ‘image enhancement’. The role of image processing functions within this data path is to improve the visual quality of the CR image in terms of spatial resolution, sharpness, contrast resolution, dynamic range, SNR. The processing efforts in the main path have to do with maximizing the information transfer to the viewer. The enhancement of image contrast is the main topic of this article and will be elaborated in section 4. The image processing operations are controlled by parameters, which often are assigned a value in accordance with the examination type. The predefined parameter values are stored in tables, with entries for each examination type. Specification of the examination type is done immediately before or after each exposure by means of an identification terminal, and hence is prone to human errors. In current CR systems some image processing functions are controlled by internal parameters which are derived from the actual image data. The parameter values are Computerized Tomography and Image Processing, 1999 DGZfP Proceedings BB 67-CD 88 Paper 16 estimated by heuristic algorithms. This way the kind and degree of enhancement are adapted to the specific characteristics of the image in terms of density levels, dynamic range, noise level, or the presence of irrelevant regions like collimation borders. In most cases a reduced version of the original image is used as input for analysis, since this still contains the essential data for the task at hand. Although the final purpose of the analysis work is very similar among different equipment, most algorithms are proprietary, so that only little is published about their actual implementation. LPDJH UHDGRXW LPDJH HQKDQFHPHQW LPDJH RXWSXW