This thesis concerns soft tissue modeling for computer-assisted medical interventions. The main objective is to take into account complex anatomical configurations combined with the influence of surgical instruments for a given medical intervention. Two particular medical procedures are studied: biopsy for prostate cancer diagnosis and brachytherapy for treatment. The aim is to give displacement and deformation information to the physician.In this thesis, a fully-discrete model is presented to simulate both anatomical and medical environnements. The elasticity function is based on a shape memory principle, allowing to simulate both rigid and elastic transformations.Needle insertion modeling is also included in the model. Experiments are proposed to assess the behaviour and accuracy of the modeling method. Comparisons of the modeling method with other soft tissue modeling methods such as the Finite Element Method and with real data are developed in order to validate the simulations. Simulations of the two targeted surgical gestures are finally detailed.