SPECT Imaging using Pinhole Collimation: System Design and Simulation Studies for Pre-Clinical and Clinical Imaging

The focus of this dissertation is the use of pinhole collimation in Nuclear Medicine. A pinhole is a single aperture in an opaque material that is placed between the detector and source of interest, and photons must pass through the hole to reach the detector. The choices of pinhole parameters, e.g. the pinhole material and size of the opening, are closely linked to the performance of the detector system.
Pinhole-based single-photon emission computed tomography (SPECT) has primarily been used to investigate human diseases in small animals because of the superior resolution and efficiency achieved with this type of collimation. The SPECT methodology involves determination of the radiopharmaceutical distribution within an object. An essential step in this methodology is the image reconstruction, i.e., the transformation of the acquired two-dimensional (2D) data into a three-dimensional (3D) distribution.
This dissertation describes the development of a SPECT system for small animal imaging called InSPECT. In Paper I, determination of the reconstructed resolution for a prototype setup with a video-based detector system is discussed, and in Paper II, construction of the InSPECT system with eight separate video detectors is described. A fusible metal, Rose’s metal, was used to cast the centre bore, in which platinum pinholes were mounted. The cast pieces could adequately shield the detectors and could be machined with a high precision. Thus, Rose’s metal, which is relatively inexpensive, could be employed for pinhole collimation but would lead to image characteristics that are less favourable than those achievable using gold or platinum pinholes. In Paper III, we present an evaluation of the performance of Rose’s metal pinholes and a comparison with other pinhole materials.
Monte Carlo-based computer models can be utilised to simulate the image formation process and can enable expensive and time-consuming instrumentation changes to be evaluated prior to building prototypes. In Paper IV, we discuss the implementation of a cadmium zinc telluride (CZT) semiconductor detector model in the Monte Carlo program SIMIND. The accuracy of the model was confirmed by comparing its results to measurements. In Paper V, we address the use of the CZT model in an SIMIND-based maximum-likelihood–expectation-maximisation iterative reconstruction of the measured data. The images reconstructed using the computer model enabled the quantification of the total activity without requiring calibration of the detector count per unit second per unit activity. In Paper VI, we describe the simulation of a dedicated SPECT scanner that employs 19 semiconductor detector–pinhole units for myocardial perfusion imaging. Compared to conventional SPECT devices, this camera is more efficient, and the efficiency is often traded for a reduction in imaging times and patient doses. An alternative use of the efficiency is to use smaller pinhole which yields better spatial resolution in the projections. No increase in contrast-to-noise was seen for smaller pinholes since any increase in contrast was also accompanied by an equal increase in noise. Also, changes in transmurality affected the CNR to a greater extent than did changes in lesion extent.