4D Reconstruction of dynamic studies with the discovery NM 530c SPECT Camera

Single photon emission computed tomography (SPECT) is a nuclear medicine imaging technique. Functional imaging allows doctors to study the physiology and function of living organs. As physiological processes in a human body are dynamic, studying changes of their temporal characteristics and spatial distribution may provide important diagnostic information. This thesis investigates various aspects of dynamic functional SPECT imaging. The algorithms are tailored for a dedicated cardiac camera system, namely the GE Discovery NM530c, a pinhole camera that can acquire views from different angles simultaneously. A fundamental feature of dynamic reconstruction is that the problem is highly underdetermined. Hence, any given consistent dataset allows infinitely many solutions. The existing dSPECT method “regularizes” the solutions by introducing constraints based on the underlying physics; it reconstructs the dynamic activity by solving one large system, as opposed to the frame-by-frame static reconstruction approach. Therefore, dSPECT reconstruction maintains temporal correlations. In this thesis, the dSPECT method was used to reconstruct images that had been scanned with the Discovery NM530c. The time activity curves of reconstructed images obtained from dSPECT are smoother than those obtained from static reconstructions. A new method, dSPECTpv, was developed to allow for successful reconstruction of dynamic behaviour that had not been observed previously in dynamic SPECT studies. Although the time activity curves of reconstructed images are much smoother, little or no improvement is observed when those reconstructions are used to estimate kinetic parameters. In this thesis, the Discovery NM530c acquisition process was modeled with the Monte Carlo method, using the simulation software GATE. Hence, computer simulations can be used to investigate the camera geometry. Our software is a useful tool in optimizing camera setup and the acquisition protocol