Abstract
This thesis is directed at the development, evaluation, and application of a novel scatter measurement technique for digital radiography. The use of an aperture for scatter measurement is examined from both a theoretical and an experimental perspective. The theoretical consideration is based on a broad beam analytical model which predicts that scatter signals should be negligible at narrow beam sizes. The experimental examination is based on analyzing features in images of narrow apertures using a digital fluoroscopy system in simulated clinical imaging conditions. It is established that a narrow aperture can generate a scatter‐free signal which can be related to the open field signal to determine the scatter signal and that this method has advantages over the conventional approach in terms of practical simplicity, accuracy, and dose efficiency. This is achieved through examination of the signature of skirt data in images of apertures 0.5 to 10 mm in diameter and through determination of the air gap dependence of aperture signals and the densitometric fidelity of image data. The approach is extended to the development and evaluation of a computerized scatter correction scheme based on two‐dimensional interpolation of scatter samples determined using an array of apertures. The evaluation is directed at the physical imaging performance of the scheme and encompasses the image acquisition and processing stages of the scatter correction process. It is established that the scheme generates substantial improvement in broad‐area contrast, densitometric linearity, square wave response factor, and spatial uniformity, has no direct effect on edge sharpness and limiting resolution, and gives rise to a substantial increase in image mottle. This conclusion is valid for any correction scheme which involves subtraction of a smooth background image be it based on spatial domain convolution filtering, interpolated scatter sampling or other process. The performance of the scheme is also compared with that of a high ratio grid and with spatial domain convolution filtering.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have