Proceedings 1999 International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine,
F. Beekman, M. Defrise, and M. Viergever, eds., pp. 21-24 (1999)

Three-dimensional reconstructions from low-count SPECT data using deformable models with interior intensity variations

Gregory S. Cunningham and Kenneth M. Hanson
Los Alamos National Laboratory


We demonstrate the reconstruction of a 3D, time-varying bolus of radiotracer from first-pass data obtained by the dynamic SPECT imager, FASTSPECT, built by the University of Arizona. The object imaged is a CardioWest Total Artificial Heart. The bolus is entirely contained in one ventricle and its associated inlet and outlet tubes. The model for the radiotracer distribution is a time-varying triangulated surface with voxel-to-voxel variations allowed inside the volume defined by the closed surface. The total curvature of the surface as well as the point-to-point variation of the interior voxel values is minimized through the use of weighted priors in the Bayesian framework. MAP estimates for the vertices, voxel values, and background count levels are produced for a subset of the 100 available 50-msec frames. The strengths of the priors (the hyperparameters) are determined by maximizing the evidence for the data over the hyperparameter values under the assumption that the posterior is approximately Gaussian.

Keywords: three-dimensional SPECT, FASTSPECT, artificial heart phantom, deformable model, variable interior density, 3D Bayesian reconstruction, smoothness prior, hyperparameters, L-curve

Get full paper (pdf, 370 KB)
Return to publication list
Send e-mail to author at