Advanced radiotherapy techniques enable highly conformal dose distribution to the tumor. This higher precision is made at the cost of an increased healthy tissue volume receiving low doses which may lead to mid and long-term adverse effects. Risk assessment requires a precise knowledge of the doses delivered to healthy tissues. However, these doses are still unknown as calculated incorrectly by the treatment planning systems (TPS).

Within this context, this thesis aims at precisely determining the doses delivered to normal tissues by advanced radiotherapy techniques. Thus, numerical and experimental tools have been developed and implemented. The Monte-Carlo model of a CyberKnife system developed in this work allows for the determination of accurate input data for risk models. Experimental developments have been used to compare the doses delivered by three radiotherapy techniques (conformational, VMAT and tomotherapy) in a pediatric renal treatment. While advanced techniques deliver highly conformal dose distribution, the doses to organs located at distance from the target are considerably increased up to a factor 3 in comparison with conformal radiotherapy. Finally, thanks to these developments, TPS capabilities have been assessed.