Skip to content
Licensed Unlicensed Requires Authentication Published by De Gruyter May 31, 2013

Improved Computed Tomography by Variable Desmearing

Model Reconstructions by Iterative DIRECTT Algorithm

Andreas Kupsch , Axel Lange , Manfred P. Hentschel and Bernd R. Müller
From the journal Materials Testing


In order to improve the spatial resolution of computed tomography reconstructions which suffer from un-sharpness due to an extended source size a new procedure of de-smearing has been developed. As the width of the source related smearing function of the projections may even double between the position near and far from the detector the desmearing procedure has to be variable. In contrast to the essentially wrong approach of de-convolution by a constant smear kernel for all voxel positions an implicit desmearing is established. The pre-known position dependent smearing is applied to the iteratively repeated projections of partial reconstructions obtained from the DIRECTT reconstruction algorithm. This approach of variable desmearing from sinograms of a pixel model with smearing from one to five pixel FWHM achieves a spatial resolution near one pixel.


Verbesserte Computertomografie durch variable Entschmierung — iterative Modellrekonstruktionen mit DIRECTT-Algorithmus. Die mitunter erhebliche Verschmierung von computertomograpischen Projektionen durch ausgedehnte Strahlenquellen wird durch eine neuartige Entschmierungsprozedur erheblich reduziert, um eine verbesserte Ortsauflösung der Rekonstruktion zu erreichen. Da die Breite der Verschmierungsfunktion der detektornahen und -fernen Objektelemente sich erheblich unterscheiden kann, muss die Entschmierung variabel sein. Im Gegensatz zu einer für alle Voxel-Positionen identischen Entfaltung, also einer grundsätzlich falschen Voraussetzung, wurde eine implizite Entschmierung entwickelt. Die vorab bekannte positionsabhängige Verschmierung wird auf die iterativ wiederholten Projektionen der Teilrekonstruktionen des DIRECTT-Algorithmus angewandt. Die variabel entschmierten Rekonstruktionen aus den Sinogrammen eines Pixelmodells mit Verschmierungen von 1 bis 5 Pixel erreichen eine Ortsauflösung von nahezu einem Pixel.

Dr. rer. nat. Andreas Kupsch, born in 1968, has studied physics at Dresden University of Technology where he made his thesis about structural properties of quasicrystals in 2004. Since then he has been engaged at the X-Ray Topography Group of the BAM (Federal Institute for Materials Research and Testing) in Berlin. His main activities refer to high energy X-ray crystallography, spectral terahertz imaging, and computed tomography.

Dipl.-Phys. Axel Lange, born in 1948, has studied physics at the Technical University Berlin where he received his M.Sc. After basic research in X-ray analysis of amorphous materials at the Freie Universität he became collaborator of the BAM, Berlin, where he has developed all basic testing equipment in founding the X-ray Topography Group. He is also the main inventor of the new CT algorithm directT.

Prof. Dr. rer. nat. Manfred P. Hentschel, born in 1943, has studied physics at the Freie Universität Berlin where he received his PhD in 1981. His postdoctoral activities have focussed on X-ray and neutron scattering of biomembranes and polymers. Since 1987 new X-ray topography techniques for nondestructive characterisation of advanced materials have been developed and applied under his leadership until his retirement from BAM in 2008.

Dr. rer. nat. Bernd R. Müller, born in 1957, has studied physics at the Technische Universität Berlin where he presented his thesis in atomic physics in 1990. After several years in basic research at several European Synchrotron Facilities he joined the X-Ray Topography Group at BAM in 1995. At the Berlin Synchrotron Laboratory BESSY he has designed and completed the BAM beamline. His activities focus on the development and application of new methods in high energy synchrotron topography and computed tomography, now leading the Topography Group.


1 F.Krejčí, J.Jakubek, J.Dammer, D.Vavřík: Enhancement of spatial resolution of roentgenographic methods. NIM A, 607 (2009) 1, pp. 208211Search in Google Scholar

2 A.Lange, M. P.Hentschel, A.Kupsch: Computertomographische Rekonstruktion mit DIRECTT: 2D-Modellrechnungen im Vergleich zur gefilterten Rückprojektion. MP Materials Testing50 (2008) 5, pp. 272277Search in Google Scholar

3 I.Manke, C.Hartnig, N.Kardjilov, A.Hilger, A.Lange, A.Kupsch, J.Banhart: Wasserverteilung in PEM-Brennstoffzellen. MP Materials Testing51 (2009) 4, pp. 219226.Search in Google Scholar

4 C.Hartnig, R.Kuhn, P.Krüger, I.Manke, N.Kardjilov, J.Goebbels, B.R.Müller, H.Riesemeier: Wassermanagement in Brennstoffzellen – die Bedeutung von hochauflösenden zerstörungsfreien Untersuchungsmethoden. MP Materials Testing50 (2008) 10, pp. 609614.Search in Google Scholar

5 J.C.Russ: The imaging processing handbock, CRC Press, New York 200610.1201/9780203881095Search in Google Scholar

6 A.Lange, M.P.Hentschel, A.Kupsch: True 3D-CT-reconstruction in comparison to the FDK-algorithm. Proceedings 17th World Conference on Non-Destructive Testing (2008)Search in Google Scholar

7 Patent DE 103 07 331 (2009)Search in Google Scholar

8 T.M.Buzug: Computed Tomography-From Photon Statistics to Modern Cone Beam CT, Springer, Berlin, Heidelberg 2008Search in Google Scholar

9 J.Radon: Über die Bestimmung von Funktionen längs gewisser Mannigfaltigkeiten. Berichte der math.-phys. Kl. Sächsischen Gesellschaft der Wissenschaften, Leipzig59 (1917) pp. 262277.Search in Google Scholar

Published Online: 2013-05-31
Published in Print: 2010-06-01

© 2010, Carl Hanser Verlag, München

Downloaded on 31.1.2023 from
Scroll Up Arrow