Thesis: At the end of the 19th century, following the discovery of X-rays by Wilhelm Röntgen, the use of ionizing radiation developed in various fields, particularly in medicine. Indeed, the appearance of interventional radiology in 1953 revolutionized patient treatment. This technique, guided by low-energy X-ray imaging (70 to 120 kV), is used for the diagnosis and treatment of various pathologies. Over the past 60 years, it has become an indispensable tool, covering more than 600 types of procedures, notably in cardiology and neurology. In France, more than 600,000 procedures are performed every year, and this number is constantly rising. Although this technique, which is less invasive than conventional surgery, is an undeniable asset for the patient and is largely under control, accidental overexposures can occur. Such accidents are rare but can lead to deterministic effects, ranging from erythema to necrosis when doses higher than 10 Gy are delivered locally to the tissue. Although the side effects most often observed with overexposure are cutaneous, the specificity of low-energy X-rays, where dose deposition is highly dependent on the composition and density of the tissue traversed, can lead to strong dose gradients. Thus, doses in dense media (bone) or adjacent to dense media can be much higher than the cutaneous dose. The lack of knowledge about the consequences and radiobiological effects of this type of exposure (low energy), where the heterogeneity of dose deposition can be significant, makes patient prognosis uncertain. This project aimed to is to contribute to the development of a new preclinical irradiation model capable of reproducing accidental overexposures in interventional radiology, to better understand the radiobiological consequences and specificities of this type of exposure through dosimetric and radiopathological characterization. To achieve this, a new model of localized low-energy (80 kV) irradiation of the paw in mice was set up on the SARRP with 5 exposure protocols (single dose: 15, 30, and 45 Gy or repeated: 2 and 3×15 Gy with 1-week intervals) and 6 post-irradiation euthanasia time points (D0 to D84). Dosimetric measurements by Electron Paramagnetic Resonance (EPR) spectroscopy on the bone at D0 and simulations highlight the high heterogeneity of dose deposition and the strong dependence on low-energy material composition. EPR measurements on the bone over time, carried out for the very first time on a preclinical in vivo model, show a loss of signal probably due to bone turnover. In the context of retrospective dosimetry following a radiological accident, dose underestimation is possible if the sample is taken several months after irradiation. Macroscopic assessment of radiation-induced lesions, using lesion scoring, weighing, and Doppler laser images to characterize blood flow, enabled us to classify irradiation protocols. Histological and microCT analyses of bone and muscle tissues showed a change in the density of irradiated muscles and a loss of trabecular bone volume at late stages. In conclusion, in this work, the new model of localized irradiation using low-energy X-rays was characterized dosimetrically and radiopathologically. It has enabled us to highlight the specific features of this type of exposure, improve our knowledge of the consequences of such exposures, and improve the prediction of the risk of complications linked to exposures in interventional radiology.