Abstract
Today, small-animal multi-pinhole single photon emission computed tomography (SPECT) can reach sub-half-millimeter image resolution. Recently we have shown that dedicated multi-pinhole collimators can also image PET tracers at sub-mm level. Simulations play a vital role in the design and optimization of such collimators. Here we propose and validate an efficient simulator that models the whole imaging chain from emitted positron to detector signal. This analytical simulator for pinhole positron emission computed tomography (ASPECT) combines analytical models for pinhole and detector response with Monte Carlo (MC)-generated kernels for positron range. Accuracy of ASPECT was validated by means of a MC simulator (MCS) that uses a kernel-based step for detector response with an angle-dependent detector kernel based on experiments. Digital phantom simulations with ASPECT and MCS converge to almost identical images. However, ASPECT converges to an equal image noise level three to four orders of magnitude faster than MCS. We conclude that ASPECT could serve as a practical tool in collimator design and iterative image reconstruction for novel multi-pinhole PET.
Export citation and abstract BibTeX RIS
General scientific summary. Small animal positron emission tomography (PET) and single photon emission computed tomography (SPECT) imaging are emerging biomedical research tools. Conventional PET is based on coincidence detection of two annihilation gamma photons, but sub-mm resolution imaging of PET tracers with pinhole collimation has been demonstrated in recent experiments. Besides high resolutions, pinhole PET has the added advantage that it provides perfectly aligned isochronous PET and SPECT images and it may save costs, floor space and imaging time because only a single device needs to be purchased. For further development of multi-pinhole collimators for PET imaging, optimization of imaging protocols and for the generation of realistic system matrices for image reconstruction, an accurate and efficient simulator of these systems is highly desirable. In this paper, a simulator that combines analytical models for pinhole and detector response with Monte Carlo generated kernels for positron range is presented and validated. This simulator converges to an equal image noise level three to four orders of magnitude faster than Monte Carlo simulations.