Next: Validation using Warped Images
Up: Experimental Validation
Previous: Image Data
In order to perform a systematic evaluation of the effects of
misregistration, we created multiple image sets, based on the MGH
Dataset, but with controlled degrees of misregistration. To
create a misregistered set, we took the original image set and
applied a set of smooth pseudo-random spatial warps, based on
biharmonic Clamped Plate Splines [20]. The warp for each
image was controlled by 25 randomly placed knot-points, each
displaced in a random direction by a distance drawn from a
Gaussian distribution whose mean controlled the degree of
misregistration introduced. This provided a very general family of
warps. We summarised the degree of misregistration by measuring
79#79, the average magnitude of pixel displacement over the whole
image. We generated a total of 70 misregistered image sets - 10
warp-set instantiations for each of 7 different values of 79#79
(0.0643, 0.249, 0.685, 1.36, 2.21, 2.76, and 4.15 pixels).
Examples of warped images are shown in Figure
.
Figure:
An example affinely-aligned brain image
and its accompanying anatomical labels, both
overlaid and expanded, for gray matter, white
matter, the lateral ventricles, and the caudate
nucleus. The labels are also divided into left and
right.
80#80
|
Figure:
An original image from the MGH Dataset (top left) and
examples of warped versions of the same image obtained using
different values of 79#79, the mean pixel displacement (shown on
each image).
81#81
|
Next: Validation using Warped Images
Up: Experimental Validation
Previous: Image Data
Roy Schestowitz
2007-03-11