Biomedical Engineering Reference
In-Depth Information
Table 2.
Optimal Values of Parameters
P
and
e
for the Minimum Mean Error
DC1
DC2
DC3
HC
P
, Pa
8000
7000
15000
8000
e
0.3
0.15
0.4
0.3
Num. Landmarks
25
21
20
21
Land. Deform., mm
2.16/3.9/1.06 1.77/2.83/0.58 1.82/3.04/0.78 4.53/6.09/0.9
Land. error, mm
1.11/2.5/0.73 1.13/2.1/0.42 1.19/2.24/0.52 1.7/3.09/0.77
Tum. vol. (true/sim.) cm
3
0.496/0.509
0.996/1.25
0.514/0.555
1.734/1.842
Optimal values of parameters
P
and
e
for MME in landmark point coordinates are reported for each
case in the used dataset. Results are for
υ
=0
.
485
. The number of landmarks, landmark deformation
statistics (mean, maximum, and standard deviation), and the model residual errors (mean, max, and
standard deviation) for the landmark points are reported. A comparison between the tumor volume in
the real images (obtained via manual segmentation) and in the simulated images is also provided.
Figure 15.
Two views of the FE mesh generated for DC1. Left: outer brain surface mesh;
right: translucent outer brain surface and colored surfaces of the tumor and the ventricles.
See attached CD for color version.
6. The deformation map resulting from each FE simulation was used to de-
form the starting image and the locations of the landmark points in this
image. Errors between the simulated deformed landmark coordinates and
rater's determined coordinates for these points in the target scan were com-
puted.
