Iterative restoration of SPECT projection images

Photon attenuation and the limited nonstationary spatial resolution of the detector can reduce both qualitative and quantitative image quality in single photon emission computed tomography (SPECT). In this paper, a reconstruction approach is described which can compensate for both of these degradations. The approach involves processing the projection data with Bellini's method for attenuation compensation followed by an iterative deconvolution technique which uses the frequency distance principle (FDP) to model the distance-dependent camera blur. Modeling of the camera blur with the FDP allows an efficient implementation using fast Fourier transform (FFT) methods. After processing of the projection data, reconstruction is performed using filtered backprojection. Simulation studies using two different brain phantoms show that this approach gives reconstructions with a favorable bias versus noise tradeoff, provides no visually undesirable noise artifacts, and requires a low computational load.