Modified sensitivity, noise equivalent count rate performance, and scatter fraction measurements of asymmetrical and dedicated brain positron emission tomographs

Abstract

Most positron emission tomography (PET) systems use an (almost) cylindrically symmetric detector geometry that acquires data in step-wise or continuous fashion. The National Electrical Manufacturers Association (NEMA) has developed performance standards (NEMA NU2-2018) to evaluate the performance of these systems. However, many Brain PET scanners no longer use a cylindrically symmetric detector arrangement; instead favoring unconventional, asymmetric spatial distributions of detectors to improve the geometric efficiency. The comparison of these systems with cylindrical devices is difficult because the NEMA standards may not be directly compatible with these non-cylindrical detector geometries. The incompatibility is due to both the source geometry and use of single-slice-rebinning (SSR). In this study, we extended the standard cylindrical polyethylene phantom used for the noise equivalent count rate (NECR) and scatter fraction (SF) measurements in NEMA NU2-2018 by adding a 20 cm diameter polyethylene sphere with a line-source channel. To avoid the use of SSR in NECR, SF and sensitivity tests, which can incorrectly assign slice locations in non-cylindrical tomographs, we instead propose a different method that uses the known positions of the line-source and the detection points of the line-of-response (LOR). Axial position can be determined from the minimum of the distance between the LOR and the line-source. These correctly binned counts were compared to cylindrical and spherical cap PET geometries using the well-validated GATE Monte Carlo code to estimate performance. The results show that our proposed modifications provide a means to estimate a non-cylindrical tomograph’s NECR, SF,and sensitivity that is consistent with the NEMA methodology.