PublicationDetail

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

THIS WEBPAGE IS DEPRECATED - please visit our new website

R. Kraus, G. Delso, H-P. Loeb, A. Martinez-Möller, R. Bundschuh, A. Drzezga, S. I. Ziegler
An inpainting approach for the correction of incomplete attenuation maps in PET/MR
Journal of Nuclear Medicine 2010, 51 (suppl. 2), p1380. Proceedings of Annual Meeting of Society of Nuclear Medicine (SNM 2010), Salt Lake City, UT, USA, June 2010 (bib)

Objectives
Attenuation correction of raw PET data is essential for reconstructing quantitative images. In whole-body PET/MR imaging, a specific MR sequence is used for the determination of the attenuation map. Unfortunately, current hybrid devices strongly restrict the field of view (FOV) of the MR. Truncated regions then lead to incomplete attenuation maps causing artifacts in the PET image. Methods: We propose to compensate these truncation artifacts in MR/PET by a geometrical inpainting approach applied to the MR image. Image structure, conceptualized in terms of level lines, is continued into the gaps following the direction of arrival at the boundary. We evaluated an algorithm that iteratively propagates smoothness information given by the discrete Laplacian. However, stronger continuation can be achieved by locally interpolating the gradient along the level lines and solving a third-order differential equation. As input, we only require missing regions to be marked in the image, e.g. by an image carving algorithm. The method was tested using coregistered MR and PET/CT datasets of the same patient that show large holes in hips, arms, and shoulders. Results: A comparable PET/CT reconstruction with equally limited FOV reveals strong artifacts, e.g. relative errors of 4.5% in the liver. Results show that inpainting effectively extrapolates missing attenuation information while preserving different tissue types. Reconstruction errors can be reduced, e.g. 1-2% in the liver, with small risk of local artifacts due to overcorrection. Large gaps (>6cm) and fine structures (<8mm) are critical and mistaken for soft tissue unless the strong continuation approach is used. Sensitivity to gap boundaries can be limited by larger masks and preprocessing the image. Conclusions: Inpainting algorithms show promise as a tool to prevent truncation artifacts in PET/MR when no additional source of patient information is available. Further work is required to include a-priori anatomical knowledge to better guide intensity propagation.</TD>
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.



Edit | Attach | Refresh | Diffs | More | Revision r1.13 - 30 Jan 2019 - 15:16 - LeslieCasas

Lehrstuhl für Computer Aided Medical Procedures & Augmented Reality    rss.gif