Effective Emission Tomography Image Reconstruction Algorithms for SPECT Data

  • Authors:

  • J. Ramírez;J. M. Górriz;M. Gómez-Río;A. Romero;R. Chaves;A. Lassl;A. Rodríguez;C. G. Puntonet;F. Theis;E. Lang

  • Affiliations:

  • Dept. of Signal Theory, Networking and Communications, University of Granada, Spain;Dept. of Signal Theory, Networking and Communications, University of Granada, Spain;Servicio de Medicina Nuclear, Hospital Universitario Virgen de las Nieves (HUVN), Granada, Spain;Dept. of Signal Theory, Networking and Communications, University of Granada, Spain;Dept. of Signal Theory, Networking and Communications, University of Granada, Spain;Dept. of Signal Theory, Networking and Communications, University of Granada, Spain;Servicio de Medicina Nuclear, Hospital Universitario Virgen de las Nieves (HUVN), Granada, Spain;Dept. of Architecture and Computer Technology, University of Granada, Spain;Max Planck Institute for Dynamics and Self-Organisation, Bernstein Center for Computational Neuroscience, Göttingen, Germany;Institut für Biophysik und physikalische Biochemie, University of Regensburg, Germany

  • Venue:
  • ICCS '08 Proceedings of the 8th international conference on Computational Science, Part I
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Medical image reconstruction from projections is computationally intensive task that demands solutions for reducing the processing delay in clinical diagnosis applications. This paper analyzes reconstruction methods combined with pre- and post-filtering for Single Photon Emission Computed Tomography (SPECT) in terms of convergence speed and image quality. The evaluation is performed by means of an image database taken from a concurrent study investigating the use of SPECT as a diagnostic tool for the early onset of Alzheimer-type dementia. Filtered backprojection (FBP) methods combined with frequency sampling 2D pre- and post-filtering provides a good trade-off between image quality and delay. Maximum likelihood expectation maximization (ML-EM) improves the quality of the reconstructed image but with a considerable increase in processing delay. To overcome this problem the ordered subsets expectation maximization (OS-EM) method is found to be an effective algorithm for reducing the computational cost with an image quality similar to ML-EM.