Thesis: Targeted therapy is based on the administration of a radiopharmaceutical, which will bind specifically in tumor regions in order to destroy them. Nowadays, this field is more and more promising thanks to the development of new radiopharmaceuticals, especially alpha emitters. Indeed, their characteristics confer a greater cytotoxicity to tumor cells while minimizing the unwanted radiation to healthy tissues in comparison with β- emitters.The goal here is to determine the administered activity, for each patient, which will ensure a maximum dose deposition in the tumor and a minimum dose deposition to the organs at risk. For alpha emitting radiopharmaceuticals, the dosimetric evaluation is a main challenge because of the particles short range.To meet this challenge, the proposed studies will focus on ²²³Ra (Xofigo®), which is the first alpha emitter that has received marketing authorization from European Commission in November 2013 for the treatment of patients with castration-resistant prostate cancer metastasized to bones. These studies were organized in three different challenges.The first challenge is to perform ²²³Ra imaging in order to determine the activity distribution in patient body. Indeed, the short path of alpha particles prevents their detection. Nevertheless, ²²³Ra and its daughters have several gamma emissions. An optimized ²²³Ra imaging protocol for gamma-camera was implemented in collaboration with the European Hospital George Pompidou. Many experiments were performed on physical phantoms. This protocol was then accepted in a new multicenter phase I/II clinical trial for the treatment of renal cell carcinoma with bone metastases.After the determination of the spatial distribution of the radiopharmaceutical, the temporal evolution must be taken into account. In order to calculate the cumulated activity from dynamic imaging, a biokinetic module has been implemented to the OEDIPE software (French acronym for “tool for personalized internal dose assessment”). This software developed by IRSN for the last fifteen years can perform precise and personalized Monte Carlo dosimetry from patient-specific anatomic and functional data.The second challenge involves the determination of the absorbed energy in the radiosensitive parts of the bone. Nowadays, the dosimetric parameters do not take the alpha particle energy, the bone site or the bone marrow proportion into account. Thus, dose calculations were optimized using the most recent and realistic bone models.Lastly, the third challenge is to characterize the distribution of ²²³Ra at the microscopic level in order to better assess the relationship between dose and biological effects. As theses parameters cannot be properly characterized on human, studies were performed on mice. Healthy mice and metastasis models, from a renal or prostate cancer, were developed in collaboration with the CIPA in Orléans. Differences of uptake between healthy tissues and metastases were studied in each model, at the microscopic scale using autoradiography methods performed in collaboration with the CRCNA at Nantes.Finally, this research work has helped to gain more insight into the various aspects of the ²²³Ra dosimetry. This work also offers tools to go further in dosimetry personalization for new alpha emitting radiopharmaceuticals, currently on the rise.