Reconstruction from Slow Rotation Dynamic SPECT Using a Factor Model

Arkadiusz Sitek, Edward V. R. Di Bella, Grant T. Gullberg

Abstract: Slow rotation acquisition of dynamic data has several advantages over fast rotation acquisition which is currently the method of choice used for the acquisition of dynamic data in SPECT. Slow rotation is currently not used because of error from inconsistent data. In this work, we develop a method of reconstructing from projections that are inconsistent in time due to being acquired during a slow acquisition. Our method is based on a factor model of physiological data. A series of dynamic images are reconstructed, where each reconstructed image corresponds in time to only one projection. Such an under-determined reconstruction is shown to be possible through utilization of a factor model. Computer simulations are performed using simple phantoms. We found that we are able to accurately reconstruct the dynamic sequence for simple phantoms with temporal behavior corresponding to teboroxime-Tc-99m heart imaging.