Tri-PET refers to high-resolution 511-keV emission tomography using a multipinhole collimator in conjunction with lower resolution PET detectors operating in coincidence mode. Tri-PET is unique in that three spatial locations are associated with each event (two detector coordinates and one pinhole location). Spatial resolution and sensitivity are similar to that of 511-keV SPECT and are governed mainly by the collimator design. However because of a third spatial location in Tri-PET, the line-of-response is overdetermined. This feature permits new opportunities in data processing which impact collimator design. In particular, multiplexing can be avoided since the coincidence data identify the pinhole through which the photon passed. In this paper, the principles of Tri-PET collimator design are reviewed and then applied to the case of high-resolution imaging of a small animal in a clinical PET scanner.Methods:
The design of a 148-pinhole collimator for whole-body imaging of a mouse is presented. Two pinhole designs were investigated: knife-edge pinholes with 1.1 mm aperture and novel hyperboloidal pinholes with 1.2 mm aperture, both having 18° cone angle. The pinhole configuration is unfocused, covering a whole-body mouse field of view with nearly uniform sensitivity. Computer simulations were performed of a micro hot rods phantom imaged with this collimator in a clinical PET scanner. Sensitivity was estimated by simulating a point source centered on-axis at locations spanning a 70-mm axial range, similar to the NEMA NU-4 standard for whole-body mouse imaging.Results:
Reconstructed images of the hot rods phantom demonstrated the ability to resolve 1.1 mm structures with the knife-edge pinholes and 1.0 mm structures with the hyperboloidal pinholes. Sensitivity was found to be 0.093% and 0.054% for the knife-edge and hyperboloidal pinholes, respectively.Conclusions:
With a properly designed multipinhole collimator, high-resolution and acceptable sensitivity are achievable with Tri-PET using ordinary clinical PET detectors.