The amount of radioactivity injected into patients during clinical PET scans can be critical when designing data acquisition protocols. The objective is to generate projection data with high statistical quality, while the acquisition time remains relatively short and the total amount of injected activity does not exceed a certain level, above which the count losses, due to deadtime effects, become significant. For this purpose an optimal range of total injected activity levels can be determined by employing the performance parameter of the Noise Equivalent Count Rate (NECR). NECR is defined as a metric of the rate in which statistically important coincidence events are counted by a PET system. The NECR depends on the total amount of injected activity and demonstrates a peak value for a certain range of activity levels. However this dependence can be affected by certain patient- and scanner-related parameters causing the shift of the range. Therefore the optimal range can be determined by estimating the NECR response as a function of the activity for a particular scanner-patient system. This is not practical for clinical studies, as it would require the repetition of the method for each patient. In this work we propose an alternative method based on a series of simulations of imaging systems and realistic human phantoms. We used Geant4 Application for Tomography Emission to simulate the independent effect of certain parameters to the NECR. We investigated the relationship between the NECR and the patient size, relative axial position of the patient to the field of view (FOV) of the scanner, combined use of reduced dead time electronics and LSO crystals instead of slowresponding electronics and BGO and finally of the energy window. We used two validated scanner models, three NCAT phantoms, two bed positions and three energy windows. The results show an optimal range of 55-65MBq for HR+ and 300450MBq for Biograph, when the heart is located at the centre of FOV.