Comparison of image Signal-to-Noise Ratio and Noise Equivalent Counts in Time-of-Flight PET

Abstract

In PET imaging, Noise Equivalent Counts (NEC) represents an estimate of the number of true coincidences necessary in an ideal system to reach the Signal-to-Noise Ratio (SNR) obtained in a real system in the presence of scatter and random coincidences. NEC has been shown to be proportional to the square of image SNR and it is used as an index of the general system performance for conventional PET images. As clinical Time-of-Flight (TOF) PET scanners are now available from all major camera manufacturers, the original NEC formula does not directly account for the contribution of TOF timing resolution to the image SNR. This study measures SNR in images of a uniform phantom for TOF and Non-TOF PET at different NEC levels to study the relationship between SNR and NEC. A cylindrical uniform phantom, 20 cm in diameter was scanned covering a range of activity levels on a Philips Gemini TF scanner. From the original scans, at each activity level, a set of data realizations were obtained through bootstrapping. NEC is calculated from the original datasets using both the original NEMA formula and alternative formulas that include the TOF gain. SNR is calculated from the bootstrapped images, reconstructed with both TOF and Non-TOF algorithms and the relationship between SNR and NEC is investigated. The results show an appreciable gain in SNR obtained by using TOF information, and a correspondence between the region where NEC and SNR peaks occur. The relationship between the two quantities, however, is not clearly established and future studies including finer sampling on the acquisition of larger phantoms and simulations are needed.