PDF version of this entire document
The overlap-based and model-based approaches were validated and compared, using a dataset consisting of 36 transaxial mid-brain slices. These were extracted at equivalent levels from a set of T1-weighted 3D MR scans of different subjects. Brain images were annotated with eight tissue classes including gray matter, white matter, the caudate nucleus and CSF (both left and right) that provided the ground truth for image correspondence. An example image and its corresponding labels are shown in Fig. 8. Initially, the images were brought into alignment using an NRR algorithm, which is based on the MDL optimisation. This alignment is assumed to be the 'correct' solution although, as a rule of thumb and as Section I explains, no registration algorithm is known to give a correct solution.