A postprocessing method for scatter compensation in single photon emission computed tomography

Single photon emission computed tomography (SPECT) is a widely used diagnostic imaging technique. The emission tomography is generated from the emitted gamma photons of a radioisotope distribution. During the acquisition of the gamma photon signals, there exist three major degradations that distort the collected data. They are attenuation, scatter and collimator blurring. Compensations for these degradations can improve the quantitative accuracy and clinical lesion detectability. This dissertation is focused on developing compensation methods for these degradations in SPECT.;An accurate attenuation correction can be achieved by either iterative or analytical reconstruction algorithms. However, the knowledge of the attenuation distribution is required. A popular way to obtain the attenuation map is to measure and reconstruct the attenuation coefficient map by using an extra CT scan. In this work, a novel method to estimate the attenuation map with the emission data only is introduced, in which the additional scan is not required. The new method is based on deriving boundaries of the constant regions of the true attenuation map using an iterative algorithm. The data consistency conditions of the attenuated Radon transform are used to select the correct attenuation map that is most consistent with the emission data.;Scattered photons degrade contrast and quantitative accuracy in SPECT images. The current approaches to compensate for the scatter are multiple energy windows subtraction methods and the iterative reconstruction-based methods. Their limitations are noise elevations and heavy computational burden. In this work, we propose a new post-processing method to compensate for the scattering effect in SPECT. Our compensation is implemented in the image domain. The scattering effect is modeled with a spatially variant two-dimensional point spread function (2D-PSF) in the reconstructed image. The 2D-PSF is estimated by using a "slab derived scatter estimation" technique. This method is firstly applied to uniformly attenuated objects and the computer simulation and phantom experiment results show an increase of quantitative accuracy and demonstrate the feasibility of this postprocessing method.;The postprocessing method is further improved and applied to nonuniform attenuation distributions. A new method for 2D-PSF estimation is proposed. The results show the restored image demonstrated an improvement in image quality with clinically acceptable computational time.