Purpose: The aim of partial breast irradiation is to reduce the risk of local recurrence into the tumor bed. After surgical resection of the tumor, the Intrabeam tm system delivers a single dose fraction of 20 Gy at the surface of the applicator, irradiating the tumor bed through a 50keV X-ray beam. However, there is currently no latitude for personalized dose prescription which is in contrast to official recommendations for individual dose optimization in radiotherapy. In this context, the study aim was to perform personalized 3D dose prediction for Intrabeam tm treatments based on patient CT imaging using Monte Carlo (MC) simulations. Methods: A model of the Intrabeam™ system was developed with GATE1 taking into account the different parts of the device2. A preoperative CT acquisition of the patient breast is performed and included in the simulation allowing accurate dose calculation. During intra-operative radiotherapy, thermoluminescent dosimeters are placed on the skin at 1 and 3 cm around the spherical applicator. Comparison between simulation results and TLD measurements are performed to confirm the dose prediction at the TLD locations. The dosimetric influence of the applicator's position is also studied on the patient CT with simulations on GATE. This study was performed on 10 patients, including 3 who had an intraoperative CT acquisition in order to validate the overall dosimetry evaluation protocol. Results: Patient results show good agreement between clinical experiments and simulations. Indeed the relative mean deviation between TLD dose measurements and GATE dose scoring is 0.09% ± 0.11% with a maximum of 0.334% and less than 1% of dose estimation uncertainty (min 0.41% max 0.95%). Conclusion: We propose the use of an accurate model of the Intrabeam system on the GATE platform based on Geant4 MC Toolkit. The dosimetric evaluation of the proposed platform with patient datasets supports its use for patient specific dosimetry planning.