Evaluation of image Signal-to-Noise Ratio in Time-of-Flight PET

In PET imaging Noise Equivalent Counts (NEC) is a common image quality index, derived to be proportional to image SNR
²
and used as an index of general system performance. Many studies have shown that TOF information leads to reduced noise, faster image convergence, and improved SNR. However, the original NEC formula does not account for any contribution of timing resolution to the final image quality, and modified versions of the formula have been proposed to account for the reduction in noise variance and increased sensitivity due to TOF information. In this study, we aim to investigate the relationship between NEC and image SNR in uniform phantoms when OSEM and TOF OSEM image reconstruction is applied. Two cylindrical uniform phantoms, 20 and 35 cm in diameter, were acquired over a wide range of activity levels on a Philips Gemini TF PET scanner. Multiple realizations of the original scans were obtained through bootstrapping and reconstructed with OSEM and TOF OSEM algorithms to obtain mean and standard deviation images. The ratio of the mean value in a central ROI over both images was taken as a measure of image SNR. NEC was calculated from the original data using both the classical and a TOF-adapted formula. The results show that Trues have a better proportionality with image SNR
²
than NEC and TOF NEC in the considered range of activities. Timing resolution and Random fraction appear to have a limited influence on image SNR, but the usage of a matched TOF kernel in reconstruction is found to be necessary to maximize the gain.