Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Restoration filtering in single photon emission computed tomographic imaging

Thesis/Dissertation ·
OSTI ID:6968945

In single photon emission computed tomography (SPECT), the reconstructions are blurred in all directions by the effects of scattered radiation, septal penetration, the collimator and detector response and the reconstruction process. Furthermore, the reconstructed slices can yield a higher fractional standard deviation than in a single planar image with the same count level. Stationary restoration filtering can be used to partially compensate for the sources of three-dimensional (3D) blur, while also suppressing noise. The restortion filter can be applied to the reconstructed data (3D post-reconstruction filtering) or to the projection images before reconstruction (two-dimensional (2D) pre-reconstruction filtering). This study describes the development of 2D and 3D Wiener filtering for use in SPECT imaging. These restoration filters were compared with a standard Butterworth smoothing filter, implemented in 1D, 2D, and 3D. Experiments were conducted to evaluate filter performance in terms of: (1) accuracy of tumor detection, and (2) ability to accurately quantitate radioisotope uptake. Simulation studies were used to test the performance of cold spot' tumor detection in the liver as determined by a mathematical observer. Results indicate that restorations obtained with 2D Wiener filtering yielded a substantially higher measure of tumor detectability than that obtained with the other filtering methods, especially the 1D Butterworth. Both restoration filters provided significantly higher lesion contrast. The investigation of activity quantitation showed that Wiener restoration filtering, combined with Bellini's attenuation correction method, can improve the accuracy of estimating radioisotope uptake over sandard Butterworth filtering.

Research Organization:
Worcester Polytechnic Inst., MA (United States)
OSTI ID:
6968945
Country of Publication:
United States
Language:
English