BrainVoyager v23.0

Fine-Tuning Alignment (FA) Step

The fine-tuning alignment step can be performed in different ways:

Gradient-driven affine transformations. This alignment approach (default) provides usually excellent results.
Boundary-based registration. This approach has been added in BrainVoyager 20.2. It produces excellent results but involves some prepatory steps to create a cortical surface mesh.
Intensity-driven multi-scale rigid-body transformations. This approach provides good results but it is more dependent on spatial inhomogeneities and does not consider scalings and shears to improve correspondence between source and target volumes.
Manual adjustment of rotation, translation and scale parameters. This approach allows to coregister FMR/DMR and VMR data sets in case that the automatic versions can not be applied, e.g. in case that only a single functional or diffusion-weighted slice has been scanned.

While the boundary-based registration is currently considered as the best coregistration method available, both gradient-driven and boundary-based registration approaches produce similar results, and are thus both recommended. The intensity-driven multi-scale approach is no longer recommended but is still available for compatibility with older versions of BrainVoyager. The three automatic FA versions are described in the next topics. The manual adjustment approach is briefly descriebed below.

Manual adjustment of rotation, translation and scale parameters

If the automatic fine-tuning does not work as desired, the coregistration might be improved by manually adjusting the rotation and translation parameters. If this step is performed after the initial alignment, the effect of changing the respective transformation values are intuitive and can be learned quickly. If this step would be performed prior to intial alignment, one would have to find large displacements and rotations making it difficult to handle this problem. Besides facilitating manual alignment in this way, QX also uses improved visualization tools to judge the alignment showing the three major orientation views for both the source and the target data set. It also allows to enter floating point values in the translation and scale parameter fields to allow obtaining optimal results.