Efficient Scatter Correction

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The last 10 years have seen significant changes in the way CT images are acquired. Commercial CT scanners have advanced from 2 to 64 slices, and the development of flat-panel thin-film transistor and other larger-area digital x-ray detectors have given rise to volumetric imaging systems with z-axis extents of 20 cm or greater. As the volume of tissue irradiated at the same time increases, so does the signal from scattered radiation that reaches the detector. The scatter signal in recorded projection images leads to a reduction in image contrast, to an increase in cupping and attenuation coefficient inaccuracy and to streak artifacts in the reconstructed CT images. Scatter-to-primary ratios (SPRs) of even a few percent produce significant streaks due to the non-linearity of the reconstruction process. The increased scatter content inherent in the image acquisition process represents a fundamental limitation of these new imaging geometries, and a new, efficient approach to scatter correction is needed. An efficient scatter correction must calculate and correct for scatter at each point in each image, should do so without requiring additional x-ray dose to the patient, should not increase the total scan time, should not add significantly to the time between data acquisition and 3D image visualization, should not introduce new artifacts, and should be widely applicable. We have developed a new, encoder-measurement-based approach for scatter correction that meets these criteria. The approach uses a primary beam modulator to encode the primary, while leaving the frequency characteristics of the scatter relatively unaffected. Simple image processing can then be used to estimate and remove the scatter signal from the projection data. We have promising results from simulations and from physical experiments on a well- characterized and stable bench-top CT system. The overall goal of this research program is to extend and optimize our approach, and to demonstrate that we can achieve an attenuation coefficient accuracy of 5 HU over a 30-cm uniform object and accuracy of 10 HU in an object with variable attenuation (ie. a clinically relevant object) for three clinical systems. Our approach has the potential to be a disruptive technology that could significantly change the way clinical systems are designed today, and could facilitate the design and translation into the clinic of new, large volume CT imaging systems.