The medical field is a vast area of application for artificial intelligence. Despite recent advances, there is still a lot of room for innovation and improvement in the face of major challenges. Deep learning, on the other hand, represents an important perspective in many areas and in particular in the field of medical imaging.In the deployment of this technique, an important restriction is related to data, its availability and confidentiality. In this thesis, we propose to offer medical experts new tools to use deep learning with a limited amount of data.We also point out the danger of pseudo-anonymization and propose a pipeline allowing true anonymization related to the patient's identity and the acquisition equipment.Le domaine médical e...