Biomedical Engineering Reference
In-Depth Information
Figure 3.
Results using the HAMMER warping algorithm. (a) Four representative sections
fromMR images of the BLSA database. (b) Representative sections from the image formed
by averaging 150 images warped by HAMMER to match the template shown in (c). (d1-
d4) 3D renderings of a representative case, its warped configuration using HAMMER, the
template, and the average of 150 warped images, respectively. The anatomical detail seen
in (b) and (d4) is indicative of the registration accuracy. The red crosses in (d3-d4) are
identically placed, in order to allow visualization of point correspondences. See attached
CD for color version.
Figure 4.
Representative example of automated definition of regions of interest, by warping
a pre-labeled atlas (left) to an individual's MR images (the warped atlas is shown on the
right as a color-coding of a volume rendering of the target brain). This automated ROI
definition makes it possible to apply the method to studies with large sample sizes in a
streamlined way. See attached CD for color version.
