This thesis presents various measurements of the difference between registrations but the lack of a gold standard registration technique did not allow us to consider the registration error in absolute terms. A registration performed using the stereotactic frame markers, such as described in Chapter 3, was shown to be more stable than a surface fit registration, but because of the problems outlined in the last section, its accuracy was not sufficient to be considered as a reference technique.
We believe however that it is possible to achieve this result by slightly modifying the OBT stereotactic frame. For example, holes in which the tip of the probe would precisely fit could be drilled at various locations on the frame. Additionnally, spherical markers could also be fixed to the frame, at accurately known positions relative to the holes, and whose coordinates could be precisely determined by finding their centroids on images. Furthermore, MRI experiments could be performed to characterize any distortion undergone by these markers. This approach would allow very accurate registrations to be performed, and absolute registration error measurement to be acheived.