Abstract
Accurate system modeling of the photon acquisition process is essential for optimizing quality in pinhole SPECT imaging. Conventional pinhole SPECT imaging assumes ideal pinhole geometry. However, neglect of pinhole finite aperture could lead to unfavorable quality degradations, such as positioning bias and image distortion. In this work, we develop a system model in which the aperture width of a pinhole collimator is explicitly included. The system model describes the probability of a single photon from its emission to detection. The probability value is calculated based on the effective intersection area resulting from a simulated cone-beam light source emitting from the image voxel, passing through a finite aperture, and reaching the detector's frontal face. The proposed model can be integrated with the ordered subsets expectation maximization (OSEM) algorithm for fast 3D statistical image reconstruction. Monte Carlo-based phantom experiments are used to evaluate the performance of the proposed system model compared to the ideal pinhole model. Reconstructed image results demonstrate that the proposed model can improve image quality in terms of reducing location bias and maintaining better contrast recovery. © 2006 Elsevier Ltd. All rights reserved.