INTRODUCTION Gliomas are neoplasms of glial cells that support and nourish the brain. ese tumors have varying histopathological features and biological behavior showing dierent aggressiveness levels, from benign-grade I to malignant-grade IV (glioblastoma multiforme). ere has been a vast amount of research in mathematical modeling to describe the growth dynamics of these tumors (Byrne et al. 2006; Cristini et al. 2003; Frieboes et al. 2007; Patel et al. 2001; Stamatakos et al. 2006; Zhang et al. 2007). Lately, specic type of macroscopic models, the reaction-diusion models, received considerable attention from the literature in the attempt to link glioma growth models to medical images (Tracqui et al. 1995; Swanson et al. 2000; Clatz et al. 2005; Jbabdi et al. 2005; Hogea et al. 2007; Mandonnet et al. 2008). ese recent models integrate information coming from medical images, specically through anatomical and diusion images, in their formulation. is integration is crucial for the transfer of mathematical models to the clinical applications since medical images are conventionally used for diagnosis and patient follow-up in the clinical routine. One of the biggest challenges in this transfer is the automatic adaptation of mathematical models to the patient, based on images. In this chapter, rst we address the problem of adapting the recent reaction-diusion models to specic patient cases using the time series of medical (magnetic resonance: MR) images (Konukoglu et al. 2009a). Following this, we address the question of retrieving relevant information for radiotherapy planning from the personalized reaction-diusion models (Konukoglu 2009).