As you can see on the main control panel (Fig. 5.1, you can switch between the different registration modes available. By default the automatic affine mode is chosen. It consists in maximizing the mutual information of the two images, and is embedded in a multi-resolution framework, based on ITK (see www.itk.org for details on the method). Three levels appear to be sufficient for a common use. One can simply click on the ``Register'' button to perform the algorithm. This operation can be executed iteratively for best results. You can aslo first choose a rigid transformation as a first step (see Sect. 5.3) to avoid failure on large transformation, and then use the affine automatic mode to get optimal results.
Note that the output of both rigid and affine registration methods is an affine transformation matrix that matches the following rules :
1. The matrix is written in homogenous coordinates in mm.
2. Considering a fixed image and a moving image , a 3D coordinate point , the output matrix can be written (in homogeneous coordinates) as follows : . One can refer to section 5.7 for further detailts on this matrix.
We also match the ITK strategy that says that a point (i,j,k) in an image is taken in the center of the the (i,j,k) voxel.