Improvement of image resolution and quantitative accuracy in clinical single photon emission computed tomography

Abstract

Clinical Single Photon Emission Computed Tomography (SPECT) is a scanning technique which acquires gamma-camera images ('projections') over a range of angles around a patient. These projections allow the reconstruction of cross sectional ('tomographic') images of the gamma-radiating pharmaceutical distribution in the patient, thus providing interesting information about the functioning of organs and tissues.SPECT images are seriously affected by a variety of image degrading processes. Restrictions on the amount of radio-pharmaceutical that can be administered to a patient cause noise in the projections and the limited spatial resolution of the gamma-camera results in blurring of the projections. In addition to these image degradations, the reconstruction of cross-sections is complicated by Compton scattering of gamma-photons in tissue, which causes attenuation of the photon flux received by the gamma-camera and causes improper detection of photons which have been scattered in tissue. This results in some additional blurring and loss of accuracy of the SPECT images in predicting activity concentrations. Tremendous efforts have been made to improve the quantitative accuracy and the spatial resolution of SPECT, and to reduce the noise in the reconstructed images. These efforts have resulted in corrective reconstruction algorithms, which are generally based on incorporation of accurate models of the main image degrading factors. Improvements of the data acquisition hardware can further increase image quality. In this paper, the image formation process of SPECT, including image-degrading factors, is explained. In addition, reconstruction algorithms and hardware modifications are reviewed, and their effects on image quality are illustrated with physical phantom and simulation experiments.