This paper describes the framework for the creation of whole-body planar acquisitions from Monte Carlo modelling with GATE. The ground truth is a complex model representing a `virtual patient' based on the NCAT-WB anthropomorphic model. Radiopharmaceutical kinetic was generated from compartmental modelling, to assign a time-activity curve (TAC) to each functional compartment. In order to match the kinetic model's functional compartments, the geometrical NCAT-WB model was used to define the corresponding functional volumes, where each geometric voxel results from a linear combination of the functional contributions for that voxel. Image generation was performed by Monte Carlo modelling with GATE. However, Monte Carlo modelling necessitates huge computation times. It is therefore important to save as much time as possible - for example by decreasing the total number of simulations required to generate a dataset. The originality of our approach is twofold: Each functional volume is simulated only once and the resulting output is weighted using the respective TAC, allowing for several image time-points to be generated from the same simulation. The second innovation deals with the way to corrupt images with Poisson noise. The classical way is to simulate enough photons to obtain almost-noiseless projections and then apply a noise filter on the GATE projections. In case of complete simulation of an anthropomorphic model, the simulation time becomes far too long. We propose a method to generate an image with sufficient statistics while preserving the Poisson noise modeled by GATE. We developed a new GATE output which stores in a list the projection pixel indices of the detected photons. It is then possible to build projections with the number of detected events needed for a specific compartment over an acquisition time period. The proposed output is compatible with the generation of SPECT images based on the compartmental approach.