In this paper, we describe the development of a new digital phantom designed for Monte Carlo simulations of breast cancer and particularly positron emission tomography (PET) imaging of the axillary lymph nodes. The phantom was based on data from the Visible Human Project female data set. The phantom covers the head-to-diaphragm regions; 17 major tissue types were segmented and 66 individual lymph nodes were identified. We have used the phantom in Monte Carlo simulations to model a reduced field-of-view PET imager based on two flat plate arrays placed on either side of the shoulder. Images used a simple single angle set of projections. We have conducted two preliminary studies: one modeling a single-frame PET acquisition 60 min after FDG injection and the other modeling a dynamic PET acquisition simulating four time frames after FDG injection. The dynamic results were processed into parametric images using the Patlak method and show the advantage to be gained by including the temporal information for legion detection. Our preliminary results indicate that the performance of such an imager forming projection images is not sufficient for axillary node PET imaging.