Noise propagation in SPECT images reconstructed using an iterative maximum-likelihood algorithm

Computer simulations were performed to investigate the effects of the photon noise in radionuclide projection data and the uncertainty in the attenuation map on the image noise in attenuation-corrected SPECT (single photon emission computed tomography) images reconstructed using a maximum-likelihood expectation-maximization algorithm. Simulated radionuclide projection data from a head phantom were acquired using a fan-beam geometry consistent with the emission-transmission (ETCT) system being developed at the University of California at San Francisco. The mean ( mu ) of the reconstructed pixel values in the region of interest started to converge after 10 iterations and the standard deviation ( sigma ) increased with the number of iterations. The sigma / mu ratio was found to be inversely proportional to the square root of the total detected counts and proportional to the relative uncertainty in the attenuation maps. These two noise components contributed independently towards the noise in the reconstructed image. In a typical ETCT system employing an X-ray tube for attenuation map acquisition, the uncertainty in the reconstructed radionuclide distribution would be limited mainly by photon noise in the projection data.<>