With the development of advanced computing and flat panel detector technology, volumetric X-ray computed tomography (CT) systems using large-area detectors, such as cone-beam CT (CBCT), are becoming more popular. This dissertation describes work aimed at reducing three types of image artifacts in CBCT.; Artifacts appear in the reconstruction when insufficient projection data are measured or the data redundancy is not properly handled. In practical CBCT systems, the circular trajectory is commonly used due to the ease of implementation on existing hardware. However, due to Tuy's data sufficiency condition, an exact reconstruction is possible only in the plane of the source trajectory. In a circular full scan, the standard FDK algorithm has been shown to be close to the optimal if unmeasured data are assumed to be zero. This algorithm usually results in cone-beam (CB) artifacts in the reconstruction, such as the intensity drop along the axial direction. An algorithm is proposed for estimating the unmeasured data. The algorithm is derived from the Radon transform and Grangeat's formula. In a circular short scan (pi plus fan angle), the modified FDK algorithm using Parker's weighting is commonly used. This algorithm handles data redundancy using approximation. Besides the axial intensity drop, the algorithm results in another type of CB artifact, mainly streaks around dense objects. An improved algorithm is derived using the central slice theorem and geometry transformation. Both the proposed full-scan and short-scan algorithms are verified using computer simulations and physical experiments.; Even if sufficient data are acquired and exact reconstruction is used, the projection data are corrupted by detected scatter, and cupping and shading artifacts appear in the reconstruction. Scatter is a large problem in CBCT systems since the geometry with a large-area detector has high scatter-to-primary ratios (SPR's). A scatter correction algorithm is introduced that provides effective scatter correction but does not require additional patient exposure. The algorithm uses a hardware-based modulation technique to separate scatter signal from the primary signal. The performance of this method is evaluated using computer simulations and preliminary physical experiments. With no loss of resolution, substantial reduction of scatter artifacts is shown using this approach.
展开▼