Accelerated 3D-OSEM image reconstruction using a Beowulf PC cluster for pinhole SPECT

ObjectiveA conventional pinhole single-photon emission computed tomography (SPECT) with a single circular orbit has limitations associated with non-uniform spatial resolution or axial blurring. Recently, we demonstrated that three-dimensional (3D) images with uniform spatial resolution and no blurring can be obtained by complete data acquired using two-circular orbit, combined with the 3D ordered subsets expectation maximization (OSEM) reconstruction method. However, a long computation time is required to obtain the reconstruction image, because of the fact that 3D-OSEM is an iterative method and two-orbit acquisition doubles the size of the projection data. To reduce the long reconstruction time, we parallelized the two-orbit pinhole 3D-OSEM reconstruction process by using a Beowulf personal computer (PC) cluster.MethodsThe Beowulf PC cluster consists of seven PCs connected to Gbit Ethernet switches. Message passing interface protocol was utilized for parallelizing the reconstruction process. The projection data in a subset are distributed to each PC. The partial image forward- and back-projected in each PC is transferred to all PCs. The current image estimate on each PC is updated after summing the partial images. The performance of parallelization on the PC cluster was evaluated using two independent projection data sets acquired by a pinhole SPECT system with two different circular orbits.ResultsParallelization using the PC cluster improved the reconstruction time with increasing number of PCs. The reconstruction time of 54 min by the single PC was decreased to 10 min when six or seven PCs were used. The speed-up factor was 5.4. The reconstruction image by the PC cluster was virtually identical with that by the single PC.ConclusionsParallelization of 3D-OSEM reconstruction for pinhole SPECT using the PC cluster can significantly reduce the computation time, whereas its implementation is simple and inexpensive.

[1]  H. Malcolm Hudson,et al.  Accelerated image reconstruction using ordered subsets of projection data , 1994, IEEE Trans. Medical Imaging.

[2]  Michel Defrise,et al.  Interest of the ordered subsets expectation maximization (OS-EM) algorithm in pinhole single-photon emission tomography reconstruction: a phantom study , 2000, European Journal of Nuclear Medicine.

[3]  H. Atkins,et al.  Pinhole SPECT: an approach to in vivo high resolution SPECT imaging in small laboratory animals. , 1994, Journal of nuclear medicine : official publication, Society of Nuclear Medicine.

[4]  H. M. Deloar,et al.  System design and development of a pinhole SPECT system for quantitative functional imaging of small animals , 2006, Annals of nuclear medicine.

[5]  L. Feldkamp,et al.  Practical cone-beam algorithm , 1984 .

[6]  Christian Kolbitsch,et al.  Evaluation of Striatal Dopamine Transporter Function in Rats by in Vivo β-[123I]CIT Pinhole SPECT , 2002, NeuroImage.

[7]  Koichi Ogawa,et al.  Ultra high resolution pinhole SPECT for small animal study , 1998 .

[8]  Y. Yonekura,et al.  Ultra-high resolution SPECT system using four pinhole collimators for small animal studies. , 1995, Journal of nuclear medicine : official publication, Society of Nuclear Medicine.

[9]  H. Iida,et al.  Absolute quantitation of regional myocardial blood flow of rats using dynamic pinhole SPECT , 2002, 2002 IEEE Nuclear Science Symposium Conference Record.

[10]  H. Tuy.  AN INVERSION FORMULA FOR CONE-BEAM RECONSTRUCTION* , 1983 .

[11]  Hiroyuki Kudo,et al.  A new reconstruction strategy for image improvement in pinhole SPECT , 2004, European Journal of Nuclear Medicine and Molecular Imaging.

[12]  M. Casey,et al.  SPMD cluster-based parallel 3D OSEM , 2002, 2002 IEEE Nuclear Science Symposium Conference Record.

[13]  Antti Sohlberg,et al.  Comparison of multi-ray and point-spread function based resolution recovery methods in pinhole SPECT reconstruction , 2006, Nuclear medicine communications.

[14]  H. Iida,et al.  Performance improvement of event-based motion correction for PET using a PC cluster , 2003, 2003 IEEE Nuclear Science Symposium. Conference Record (IEEE Cat. No.03CH37515).

[15]  L. Shepp,et al.  Maximum Likelihood Reconstruction for Emission Tomography , 1983, IEEE Transactions on Medical Imaging.

[16]  Ma Tianyu,et al.  Parallel OSEM Reconstruction Algorithm for Fully 3-D SPECT on a Beowulf Cluster , 2005, 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference.

[17]  Ulla Ruotsalainen,et al.  Accelerated median root prior reconstruction for pinhole single-photon emission tomography (SPET). , 2003, Physics in medicine and biology.

[18]  R. Jaszczak,et al.  Pinhole collimation for ultra-high-resolution, small-field-of-view SPECT. , 1994, Physics in medicine and biology.

[19]  T. Hirai,et al.  Serial evaluation of fatty acid metabolism in rats with myocardial infarction by pinhole SPECT , 2001, Journal of nuclear cardiology : official publication of the American Society of Nuclear Cardiology.

[20]  C. Michel,et al.  Impact of a high-performance communication network on cluster-based parallel iterative reconstruction , 2005, IEEE Nuclear Science Symposium Conference Record, 2005.

[21]  R M Leahy,et al.  Internet2-based 3D PET image reconstruction using a PC cluster. , 2002, Physics in medicine and biology.

[22]  Michael E. Casey,et al.  HeinzelCluster: accelerated reconstruction for FORE and OSEM3D. , 2002 .

[23]  P. Acton,et al.  Quantification of dopamine transporters in the mouse brain using ultra-high resolution single-photon emission tomography , 2002, European Journal of Nuclear Medicine and Molecular Imaging.