Adam M. Alessio PhD, Presenter: Research Grant, General Electric Company Consultant, Lantheus Medical Imaging, Inc

Aaron So PhD, Abstract Co-Author: Nothing to Disclose

Ting-Yim Lee MSc, PhD, Abstract Co-Author: Grant, General Electric Company Royalties, General Electric Company

Partial scan (PS) CT acquisitions lead to artifacts due to inconsistencies from scatter contributions and inexact cone-beam reconstruction. Dynamic cardiac studies, which benefit from PS acquisitions and require high CT number fidelity, require corrections for these view-angle dependent shading artifacts. PS artifact reduction methods have been proposed based on raw projections, images, or combinations of projection and temporal information. We propose a PS artifact reduction (PSAR) method requiring only access to images to reduce the influence of cone-beam and scatter induced artifacts.

The cone-beam issues are mitigated with an error term derived from the error-reduction-based algorithm (ERB, Zeng et al, 2004). In addition, the scatter contribution is estimated based on the convolution of pencil-beam scatter distribution and a forward scatter model. These two components are iteratively estimated through successive cone-beam forward projections and Feldkamp reconstructions of the original artifact-present images. The proposed PSAR algorithm was tested with simulations of a 64-slice CT acquisition of a thoracic phantom. The view-angle dependent error was evaluated in simulations and verified with measured images from DCE-CT porcine studies.

Simulation studies revealed that for a realistic thoracic morphology, CT numbers in slices at the edge of the axial field of view (from large cone-angles) vary by 30 HU as a function of the center view angle of the PS acquisition. Scatter alone causes variations of ~6 HU at different view angles. PS’s from porcine studies confirmed HU variations in soft-tissue regions of 29 HU as view angle varied. In simulations, the application of the PSAR algorithm reduced the RMSE in the myocardial region from 14.5 HU to 6.3 HU across all slices and view angles.

PSAR reduced view-angle dependent artifacts in PS acquisitions. This method benefits from not requiring access to raw projection data or full-scan information, offering a practical method for artifact reduction in DCE-CT studies.

The proposed PSAR method reduces scatter and cone-beam artifacts in partial scan acquisitions to improve the quantitative information from these fast, lower-dose dynamic cardiac CT acquisitions.

Alessio, A, So, A, Lee, T, Image-based Partial Scan Artifact Reduction for Dynamic Contrast Enhanced CT.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13029053.html