Développement d'algorithmes temps réel de traitement de séquences d'images animées biomédicales pour la thermothérapie guidée par IRM

During the last years MR-guided lo cal hyp erthermia has b een develop edfor the clinical practice. Hyp erthermia is based on the destruction of tumorcells due to an increased temp erature over a sufficient amount of time. The si-multaneous acquisition of MR images during the heating pro cess allows on-linemonitoring of the temp erature and a control of the tissue destruction inducedby the hyp erthermia treatment. The on-line calculation of the temp erature isnot a trivial problem. This is b ecause the physical and physiological pheno-mena like the respiration disturb the temp erature measurement. Hence, it isnecessary to develop strategies for real-time correction in order to deliver pre-cise temp erature maps. In this work, we prop ose several algorithms for imagetreatment enabling the correction of artefacts related to motion and magneticsusceptibility effects. These corrections will provide a reliable estimation of theefficiency of the hyp erthermia treatment.