Evaluation of stationary and semi-stationary acquisitions in a dual-head multi-pinhole system for myocardial perfusion SPECT

Abstract

Previously we proposed a multi-pinhole (MPH) collimator for stationary clinical myocardial perfusion imaging (MPI) based on a dual-head clinical SPECT/CT scanner. However, artifacts from non-sufficient angular sampling were observed. This study aims to investigate different stationary MPH designs and the potential improvement in image quality by semi-stationary configurations with increased number of acquisition views. The 2 detectors were connected at ∼90° or L-mode with minimal imaging distance of 18.3 cm for MPI. We first optimized the MPH collimator parameters including pinhole number, imaging distance, collimator length, aperture size and acceptance angle by maximizing the sensitivity for a pre-determined field-of-view of 16 cm and target resolution of 1.095 cm. Based on the optimized results, a 3D analytical MPH projector was used to generate noise-free projections of (i) five stationary designs (L-mode-I to L-mode-V) and (ii) semi-stationary designs with 4 detector positions using the 4D extended cardiac torso (XCAT) phantom which was scaled to fit for 95% of Asian male population. Simulations of low energy high resolution (LEHR) collimator was used as a reference. The angle between 2 L-mode positions in the semi-stationary acquisition was evaluated from −45° to 180°. A 3D MPH ML-EM/LEHR OS-EM algorithm was used to reconstruct the MPH/LEHR data and normalized-mean-square-error (NMSE) was evaluated in a 2D ROI with 32 × 32 pixels which covered the center slice of the left ventricle. Our optimized MPH parameters are 12-pinhole, acceptance angle of 51°, aperture size of 4.65 mm, collimator length of 15.8 cm and imaging distance of 18.6 cm. The NMSE of L-mode-I was the lowest (0.3021) among all stationary designs. Semi-stationary mode with 40° rotation based on the stationary L-mode-I configuration provided the least NMSE result (0.1306) among all acquisitions which was better than the LEHR (0.2186). We concluded that semi-stationary MPH acquisition substantially improved the image quality as compared to the full stationary design as well as the conventional LEHR in MPI.