Cardiac CT can be achieved by performing short scans with prospective gating. As the collimation of multi–slice CT scanners generally does not allow for a coverage of the entire heart, sequence scans, also known as step-andshoot, can be used, where irradiation is performed multiple times for varying positions. Each of these short scans yields data, generally with a longitudinal overlap, that can be reconstructed into a sub-volume, or stack. The latter ideally corresponds to the same phase. The issue addressed in this work is irregular motion, such as irregular heart motion. It leads to stacks that do not represent exactly the same volume, resulting in discontinuities at stack transitions when assembling the complete CT volume. We propose a stack transition artifact removal method including a simple symmetric registration approach. Originating from a set of control points in overlap regions between adjacent stacks, the algorithm symmetrically searches for matching sub volumes in the two neighboring stacks, respectively. The offsets to the respective control points of matching sub volumes is used to compute two deformation vector fields that match the two stacks to each other. The deformation vector fields are extended from the overlapping regions in order to maintain smooth and anatomically meaningful images. We validated the method using clinical data sets. By applying a straightforward symmetric registration method to cardiac data, we show that the stack transition artifacts can be addressed in this fashion. The artifact removal was able to considerably improve image quality and constitutes a framework that can be enhanced and expanded on in future.
Edge–preserving adaptive filtering within CT image reconstruction is a powerful method to reduce image noise and hence to reduce patient dose. However, highly sophisticated adaptive filters typically comprise many parameters which must be adjusted carefully in order to obtain optimal filter performance and to avoid artifacts caused by the filter. In this work we applied an anisotropic tensor–based adaptive image filter (TBAF) to CT image reconstruction, both as an image–based post–processing step, as well as a regularization step within an iterative reconstruction. The TBAF is a generalization of the filter of reference.1 Provided that the image noise (i.e. the variance) of the original image is known for each voxel, we adjust all filter parameters automatically. Hence, the TBAF can be applied to any individual CT dataset without user interaction. This is a crucial feature for a possible application in clinical routine. The TBAF is compared to a well–established adaptive bilateral filter using the same noise adjustment. Although the differences between both filters are subtle, edges and local structures emerge more clearly in the TBAF filtered images while anatomical details are less affected than by the bilateral filter.
The reconstruction of CT images with low noise and highest spatial resolution is a challenging task. Usually, a trade-off between at least these two demands has to be found or several reconstructions with mutually exclusive properties, i.e. either low noise or high spatial resolution, have to be performed. Iterative reconstruction methods might be suitable tools to overcome these limitations and provide images of highest diagnostic quality with formerly mutually exclusive image properties. While image quality metrics like the modulation transfer function (MTF) or the point spread function (PSF) are well-defined in case of standard reconstructions, e.g. filtered backprojection, the iterative algorithms lack these metrics. To overcome this issue alternate methodologies like the model observers have been proposed recently to allow a quantification of a usually task-dependent image quality metric.1 As an alternative we recently proposed an iterative reconstruction method, the alpha-image reconstruction (AIR), providing well-defined image quality metrics on a per-voxel basis.2 In particular, the AIR algorithm seeks to find weighting images, the alpha-images, that are used to blend between basis images with mutually exclusive image properties. The result is an image with highest diagnostic quality that provides a high spatial resolution and a low noise level. As the estimation of the alpha-images is computationally demanding we herein aim at optimizing this process and highlight the favorable properties of AIR using patient measurements.