The Techa River (the Urals, Russia) was heavily contaminated due to the release of radionuclides from the Mayak Production Association. The radioactive releases included bone-seeking beta-emitters such as 90 Sr and 89 Sr that contribute to doses to bone marrow. Moreover, 90 Sr is a long-lived isotope, the uptake of which leads to a chronic bone marrow exposure known to result in an increased risk of leukemias. Ongoing epidemiological studies of the long-term effects of chronic radiation exposure are being performed for the Techa River Cohort members. Radiation dosimetry is a part of this study, of which, the internal dose estimates for active bone marrow exposed to beta emission of Sr isotopes incorporated in calcified tissue is an important component. Internal dose calculations, which are based on electron-photon transport simulations, require geometrical descriptions of bone shapes and bone microstructures of the main hematopoietic sites of the human skeleton (ribs, vertebrae, pelvic bones, femur, humeri, bones of the skull, sternum, clavicle and scapula). For this purpose, the parametric approach for modeling bone geometry was elaborated. The proposed approach can be used to segment and define each of the bone sites as simple geometric shapes for which parameters can be derived. The aim of the paper is to present the principles of bone segmentation. Dose factors that convert the activity concentration of bone-seeking radionuclides into a corresponding bone marrow dose rate were calculated for each segment. The calculations were done with MCNP6. The bone segmentation allows optimizing the size of the computational phantom (in terms of voxel number), substituting the complex-shaped bone model with a set of simple stylized phantoms and taking into account the heterogeneity of bone microstructure.