Correction of organ motion in SPECT using reprojection data

Abstract

Organ motion during SPECT acquisition can cause artifacts in the reconstructed slices which may affect the clinical interpretation. A new algorithm is presented which corrects for translational organ motion in SPECT projection data and thus reduces the resulting artifacts. The projection data is first reconstructed and then reprojected to the original angles. These reprojection data provide a template for comparison with the true projection data to estimate motion. An organ of interest is selected interactively using a count threshold on the reconstructed slices. From this, regions of interest are reprojected and used to limit the region of comparison and improve efficiency. Parabolic interpolation is used to interpolate fractional shifts. The algorithm corrects for translational motion in the projection data along both X and Y axes. The algorithm was evaluated using point sources in air with physical motion induced during acquisition and artificially induced motion using Monte Carlo simulated MCAT phantom data. Results show that the FWHM's and FWTM's of corrected point source data compare favorably with point source data without motion induced. Polar maps of the MCAT phantom clearly show the improvement after the correction was implemented. Further validation and clinical testing are in progress.