Subthalamic nucleus (STN) Deep brain stimulation (DBS) has demonstrated its efficacy in the treatment of Parkinson´s disease. Accurate targeting of the STN is a crucial step for the success of DBS surgery. MRI has shown to be reliable and allows direct targeting. Unfortunately, STN is not always clearly identifiable on standard T2-weighted images. In these situations, manual targeting procedures represent only an estimation of the real STN position. Recently, registration algorithms have been developed to automatically estimate the position of brain structures from segmented electronic atlases. However, their precision remains to be studied. We propose to evaluate the performance of different targeting methods (manual and automatic) compared to expert accuracy.