TITLE: Theoretical Implications and Design Principles of Focused Multi-Pinhole SPECT


AUTHORS: Yuchuan Wang, David Graff, Greta Mok, Benjamin Tsui


PURPOSE: We study, in theory, the advantages and implementations of focused multi-pinhole (FMPH) SPECT, in which, all of the pinhole apertures are tilted towards a common rodent-sized field-of-view (FOV) for maximum gain in detection efficiency.


METHODS AND MATERIALS: First, we establish two equations using system design parameters: one describes the photon detection efficiency (PDE), the other describes how well the Tuy's condition is satisfied within the FOV (defined as TSE: tomographic sampling efficiency). Among "resolution-equivalent" system designs with minimal projection-overlap, we show that the number of unknowns in the PDE equation is reduced to two: the number of pinholes (N) and the system radius-of-rotation (R) while maintaining a non-truncated FOV.

Next, we analyze that an optimal FMPH SPECT system should have high values of PDE and TSE. Therefore, designing such an imaging system becomes a process of finding system parameters that maximize PDE, and then TSE. Parameters N and R are directly obtained from PDE maximization, while aperture diameter (D) and collimator length (F) are calculated based on them. We then analyze the factors that affect TSE, including the number/placement of the pinholes and the total number of angular views. This analysis guides the placement of N pinholes, which completes the design process. Implications of PDE and TSE maximization are shown by comparing FMPH SPECT and single pinhole (SPH) SPECT.


RESULTS: For a typical rodent imaging configuration of 26mm FOV and 25mm minimum R, to achieve a target resolution of 1.2mm, PDE maximization for a clinical SPECT camera (40cm detector size, 3.5mm intrinsic resolution) resulted in N=8, R=25mm, D=0.8mm, F=12 cm, and 4 times PDE over a resolution-equivalent optimal SPH design. For a modular camera (12cm detector, 1.5mm intrinsic), we obtain N=4, R=25mm, D=0.6mm, F=5cm, and 1.8 times PDE over the optimal SPH design.

To maximize TSE, we found that one should minimize the use of collinear sub-patterns, especially those parallel or perpendicular to the scanning orbit. Together with the consideration of minimizing projection overlap, we obtained two practical pinhole patterns for the above two cases, which have a TSE gain of 90% and 58% over SPH designs, respectively.

Simulation and experiment studies using the proposed FMPH SPECT designs demonstrated superior reconstructed image quality throughout the entire FOV as compared to the corresponding SPH SPECT. The benefit may also be traded for significantly reduced acquisition time without sacrificing reconstructed image quality.



CONCLUSIONS: FMPH SPECT may be a superior imaging method as compared to SPH SPECT in practical rodent imaging studies. We developed effective means of optimizing the FMPH design based on maximizing both photon detection efficiency and tomographic sampling efficiency.