Volumetric sampling, analysis, and visualization in spiral CT

Efforts to improve the clinical utility of medical data are usually complicated by the limited resolution available in the dataset, and by artifacts in the reconstructed images. The introduction of spiral CT has enabled an increased rate of coverage and the potential for improved longitudinal resolution in the volume. Both of these factors point to the need for improved visualization techniques to handle the increased amount of data, and to take advantage of the improved resolution in the volume.; In this thesis, we present an analysis of volumetric sampling in spiral CT to investigate, in particular, the artifacts caused by inadequate sampling in the data acquisition process. We propose a mathematical model to describe the three-dimensional (3D) sampling scheme in spiral CT. Based on this model, we propose a theory to explain the spatially varying longitudinal aliasing in spiral CT. This longitudinal aliasing is negligible at the isocenter, but can become significant at locations far from the isocenter. In addition, the longitudinal aliasing in spiral CT is a function of the table speed, and the helical interpolation algorithm used in the reconstruction process.; We investigate the effects of this spatially varying longitudinal aliasing on longitudinal resolution. Our results show that the distortions caused by the aliasing can severely affect longitudinal resolution at locations far from the isocenter. This effect is more pronounced at higher pitch values. We propose a new measure to quantify the effects of aliasing on resolution, the contrast-to-aliased-noise ratio (CN_aR), which treats aliasing as a signal dependent and additive noise. CN _aR values are useful for estimating the amount of distortion present for different sets of scanning parameters and at different transaxial positions. The impact of the distorted longitudinal resolution on clinical applications such as lesion detection was found to be diagnostically significant.; A novel volume visualization technique, the sliding-thin-slab volume visualization (STS-VV), is also proposed. This visualization approach allows interactive exploration of complex medical volumes by rendering thin slabs at orientations and positions within the volume specified by the user. This approach renders views of anatomical structures without occlusions and with improved conspicuity.