Computational Medicine efforts related to translating Cancer computational models to the clinical practice are focusing on identifying and testing ways to validate the models proposed in vivo before tumor resection. In real life this is actually difficult if not impossible, since patients are treated right away and there is no direct way of imaging the tumor growing. However, in this work, we attempt to validate the simulated outcome of a mathematical tumor growth model of reaction diffusion type with the actual tumor behavior in human cancer cell lines injected subcutaneously and grown as xenografts in immunodeficient mice by utilizing fluorescence molecular tomography performed in vivo. We show that knowing the initial spatial concentration of the viable cancer cell population, as well as hypoxia and vascularity significantly improves the in silico predictions. Such simulations provide patient specific details that play a significant role in the evolution of the tumor under study.