Brian Edward Nett PhD, Presenter: Employee, General Electric Company

Zhye Yin, Abstract Co-Author: Employee, General Electric Company

Jed Pack, Abstract Co-Author: Employee, General Electric Company

To assess the performance of a new image reconstruction algorithm, which properly weights redundant data, for the case of a half-scan type temporal weighting function. 

The most common method to achieve half-scan type temporal weighting for cone-beam tomography is to extend the standard Parker type weighting function which accounts for redundancies in sinogram space (P-FDK). This weighting is exact in the central slice but leads to data mishandling away from the central slice. More recent methods have been developed which properly account for redundant data (e.g. application of Katsevich framework for exact reconstruction to a single arc source trajectory). In this work a method which properly weights redundant frequency data is evaluated for repeated quantitative imaging. The effect of the reconstruction algorithm is studied in isolation from other physical effects by using a monoenergetic simulation of an anthropomorphic phantom (ie. the XCAT phantom). The central view angle of the image reconstruction was sampled in 15 deg intervals over 360 deg.

The standard deviation of reconstruction values as a function of central view angle was computed in order to visually demonstrate the strong angular dependence of standard methods. Automated regions of interest (ROIs) were defined for myocardium and fat tissue types in the phantom and measured as a function of cone-angle. For cone angles greater than +/-2 deg the percentage of voxels which had a std. w.r.t. central view angle of greater than 10 HU was [P-FDK] (96.9%,96.4%) , [improved recon] (0.14%,1.4%) for myocardium and fat ROIs respectively.

New image reconstruction methods indicate that repeatable quantitative imaging is possible for CT acquisitions with a large cone angle.

Advanced image reconstruction techniques can improve repeatable quantitative reconstruction for half-scan reconstruction.

Nett, B, Yin, Z, Pack, J, Proper Weighting of Redundant Data in Cardiac Reconstruction Enables Repeatable Quantitative Imaging at Large Cone Angles.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13016160.html