Multidetector-Row CT: Image-Processing Techniques and Clinical Applications

Abstract

Image processing and three-dimensional (3D) reconstruction of diagnostic images represents a necessary tool for depicting complex anatomical structures and understanding pathological changes in terms of both morphology and function. The importance of 3D reconstructions is evident if we consider that the quantity of native images produced with new-generation cross-sectional techniques has become increasingly large. Volumetric data such as those acquired with multidetector row computed tomography (CT) are particularly well suited to postprocessing. On the other hand the analysis and processing of such data through additional planes over the axial and 3D views is becoming mandatory. Image processing involves operations such as reformatting original CT images and surface and volume rendering. These types of operations are also included in a wide classification which divides the techniques of display of 3D models into projectional and perspective methods. Projectional methods are those in which a 3D volume is projected into a bidimensional plane; in the perspective methods a 3D virtual world is displayed by means of techniques that aim to reproduce the perspective of the human eye looking at the physical world. Projectional methods include CT image-reformatting approaches such as multiplanar reformations (MPR) in the sagittal, coronal, oblique, and curved planes. More specific projection techniques include maximum-intensity projection (MIP) and minimum-intensity projection (MinIP). The reformatting process does not modify the CT data but uses them in off-axis views and displays the images in an orientation different from native acquisition. Surface and volume rendering use algorithms that generate 3D views of sectional two-dimensional data. Surface rendering is based on the extraction of an intermediate surface description of the relevant objects from the volume data, while volume rendering displays the entire volume preserving the whole dynamic range of the image. A more advanced application of surface and volume rendering is represented by virtual endoscopy, which is a simulation of the endoscopic perspective by processing volumetric data sets. The CT acquisition parameters that have a direct effect on the quality of the image processing are section thickness, reconstruction spacing, and pitch. Thin sections and reconstruction spacing allow better postprocessing results by reducing partial volume averaging effects on the longitudinal plane or z-axis (Fig. 3.1). The effect of pitch on 3D imaging of CT data sets is particularly relevant for single-row systems, where the use of high pitch values introduces an increased slice-sensitive profile and consequently determines artifacts on projectional and perspective CONTENTS