Spatial Normalization of Image Data

In our discussion up to now, we have described how the morphology of an individual brain can be mathematically represented by adapting a template to the shape of that brain. Very often in the brain imaging literature researchers use a dual technique for analyzing image data. In particular, they spatially transform image data to a common reference system, which is associated with a template. This transformation is actually the inverse of the transformation we discussed earlier: It takes a point in the individual's brain and maps it to a point in the template brain, which resides in a common reference space often called a stereotaxic space. Image data mapped this way can be structural images, in which the various kinds of brain tissues have been labeled according to some segmentation methodology [34], or functional images, such as positron emission tomography (PET) or functional magnetic resonance imaging (fMRI), that reflect activation of the brain during certain tasks. We will treat these two separately.

Depth for Subject AC

Position along the sulcus

FIGURE 8 (a) An example of a sulcal ribbon overlaid on which are shown several depth probes found via a dynamic programming algorithm [33]. Roughly, these probes represent paths along the surface that connect the outer edge of the sulcus with the deepest edge of the sulcus, and that are as close as possible to a plane that is normal to the ribbon. (b) The resulting depth measurements along the sulcus, as a function of arc length.

Depth for Subject AC

Position along the sulcus

FIGURE 8 (a) An example of a sulcal ribbon overlaid on which are shown several depth probes found via a dynamic programming algorithm [33]. Roughly, these probes represent paths along the surface that connect the outer edge of the sulcus with the deepest edge of the sulcus, and that are as close as possible to a plane that is normal to the ribbon. (b) The resulting depth measurements along the sulcus, as a function of arc length.

0 0

Post a comment