Biomedical Engineering Reference
In-Depth Information
difference images and calculated statistics such as the voxel mean and standard
deviation following registration.
Finally, we used a variety of tools in RegViz to evaluate registration quality.
We used contour overlap and color overlay to assess the prostate registration. We
manually segmented the prostate across all slices and calculated the potential
displacements of the prostate 3D centroid.
3.3.4
Examples and Results
3.3.4.1
Image Acquisition
All MRI volumes were acquired using a 1.5 T Siemens MRI system (Magne-
tom Symphony, Siemens Medical Systems, Erlangen, Germany). As described
in Section 3.2.4.1, we used two MR sequences. First, a 3D FLASH sequence
with TR/TE/flip parameters of 12/5.0/60 gave 256
×
256
×
128 voxels over a
330
×
330
×
256-mm field of view (FOV) to yield 1
.
29
×
1
.
29
×
2
.
0-mm voxels
oriented to give the highest resolution for transverse slices. This sequence was
used for volunteer S1. Second, a 3D PSIF sequence with 9.4/5.0/60 (TR/TE/flip)
yielded 160
×
256
×
128 voxels over a 219
×
350
×
192-mm rectangular FOV and
1
.
4
×
1
.
4
×
1
.
5-mm voxels oriented to give the highest resolution for transverse
slices. The second sequence was used for volunteers S2 and S3.
3.3.4.2
Imaging Experiments
We acquired 3D MRI volume images from three normal volunteers under a va-
riety of conditions simulating anticipated conditions in diagnostic and treat-
ment applications. Before image acquisition, each volunteer drank water and
had a relatively
full bladder
.Inthe
diagnostic position,
the subject laid supine
throughout MR scanning. In the
treatment position,
the subject was supine, and
his legs were supported at 30
◦
-60
◦
relative to the horizon. In some experiments,
the subject micturated to create an
empty bladder
prior to imaging. We imaged
volunteers a week before the standard imaging session, and we refer to these
volumes as
diagnosis 1 week
. Between volume acquisitions, volunteers got off
the MRI table, stretched, and walked around to ensure that they would assume
a different position when they laid back on the table. All images of a volunteer
were acquired with the same MRI acquisition parameters so as to ensure very