Michael Meyer, Abstract Co-Author: Nothing to Disclose

Yiannis Kyriakou PhD, Presenter: Nothing to Disclose

Wei Chen MEng, Dipl Eng, Abstract Co-Author: Nothing to Disclose

Willi A. Kalender PhD, Abstract Co-Author: Consultant, Siemens AG Consultant, Bayer AG Founder, CT Imaging GmbH Scientific Advisor, CT Imaging GmbH Shareholder, CT Imaging GmbH Founder, Artemis Imaging GmbH CEO, Artemis Imaging GmbH Shareholder, Artemis Imaging GmbH

In flat-detector CT (FD-CT) scatter fractions increase with increasing exposure volume and cause cupping or streak-like artifacts in the reconstructed images. Our goal was to develop an efficient scatter artifact reduction without the need for a priori knowledge.

The method is based on observations of scatter properties in both the image and projection domain. The algorithm is independent of scanner geometry and of acquisition parameters such as spectrum or detector. Within the image domain, cupping is assessed by fitting of a 3D-polynomial P, followed by forward projection of two water-equivalent objects with and without added P. Subtraction of both projections yields a first scatter distribution estimate. As scatter distributions are usually flat for typical objects, several smoothing versions were tested. Reconstruction of the corrected projections is then assessed for remaining scatter artifacts and the scatter levels are iteratively adjusted to reach a target value for mean central soft tissue. The method was evaluated in simulations and measurements. For simulations, the root-mean-square value of the difference (RMSD) between the CT values of the reconstructed image and an ideal scatter-free image was calculated. The difference of the RMSDs for the uncorrected and the corrected case divided by the RMSD of the uncorrected case was used as a figure of merit Q. In addition, cupping, contrast and contrast-to-noise ratio (CNR) was evaluated. The algorithm was evaluated on commercial systems including truncated and non-homogeneous clinically relevant objects.

For clinically relevant objects, the CI version, which calculates and stores just one mean scatter value per projection, performed best. It significantly reduced cupping and increased CT value accuracy in both simulations and measurements. For simulated voxelized head, thorax and hip phantoms, a Q of 0.73, 0.74 and 0.76 was obtained, respectively. For a CatPhan measurement, cupping was reduced from 12.2% to 2.2% and CT value accuracy was increased for all inserts.

The presented method provides a fast and practical reduction of scatter artifacts for FD-CT and improves image quality without any a priori knowledge.

Scatter artifacts in FD-CT are reduced significantly without impacting routine workflow.

Meyer, M, Kyriakou, Y, Chen, W, Kalender, W, Heuristic Scatter Artifact Reduction in Cone Beam CT.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9011457.html