Evaluation of clinical PET count rate performance

We propose a methodology for analyzing clinical PET count rate performance that involves matching net trues (prompts - delayeds) and randoms (delayeds) count rate responses measured on a reference phantom to the actual patient data. We employ these estimated response curves to compute a performance metric closely related to the noise equivalent count rate, as a function of the total singles event rate in the system. Front this we can determine the peak performance value relative to the measured performance for any individual acquisition. This maximum performance value is largely independent of the magnitude of the activity present in the patient at the time of the acquisition, but depends mainly on the emitter and attenuator distributions. These peak count rates can be used to derive frame durations for equivalent noise, and correlated with patient weight. We can also determine lite singles rate, and thereby activity, necessary to achieve maximum performance, and based on the known activity in the patient, predict optimal injected doses.