Information Technology Reference
In-Depth Information
adolescent kyphoscoliosis reduces the risk of cancer in these patients, as ultrasound
has no known adverse side effects. Tracked ultrasound machines are also more
accessible tools than X-ray machines. Portable ultrasound machines allow screen-
ing and monitoring in remote areas where permanent medical imaging facilities are
not available. Therefore, tracked ultrasound may become the clinical standard for
kyphoscoliosis monitoring in the future.
16 Ultrasound Image Fusion with Other Modalities
Ultrasound imaging lacks important features of CT or MRI modalities, including
characteristic image intensity values for different tissues. Intensity values are rel-
ative on ultrasound due to attenuation, acoustic shadowing, and other artifacts. The
ideal image guidance for the interventionist would have the standard image quality
of CT and MRI, and also the convenient accessibility of ultrasound. Therefore, a
great challenge for researchers and engineers is to fuse preoperative CT and MRI
with ultrasound in real time during ultrasound scanning. If these preoperative
images are registered to the patient anatomy, tracked ultrasound images can be
enhanced by showing a corresponding slice from CT or MRI, either fused with the
ultrasound, or side-by-side. Tracking ensures that both images show the same slice
respective to the patient anatomy. Even though perfect spatial registration between
preoperative images and intraoperative ultrasound cannot be achieved due to soft
tissue deformations around the spine, and due to patient motion, physicians can
mentally correct for these deformations, so the image fusion can help both even
when the registration accuracy is limited.
CT-to-ultrasound or MRI-to-ultrasound fusion could also be used to eliminate
needle tracking from interventional procedures. Ultrasound can be used to directly
visualize the needle, and preoperative images show the target anatomical structures.
Therefore, fusion of the two modalities may provide real time needle navigation in
preoperative images. This potentially reduces the cost of disposable needle trackers,
and extends the applicability of tracked ultrasound to interventional tools (e.g.
tissue ablators) that are currently not equipped with position tracking.
Signi
cant effort has been made to implement fusion of preoperative images
with intraoperative ultrasound. The registration methods are either based on com-
mon image features between CT and ultrasound [
27
], or they use the surface model
of the spine, which requires segmentation of the vertebrae [
28
]. A common problem
in image registration is that the CT image is usually taken in supine patient position,
while needle insertions are done while the patient is bent forward. This requires
non-rigid registration of the CT image. Biomechanical constraints can be applied to
account for the typical deformation of the spinal column. Unfortunately, rate of
failed spine CT to ultrasound registration is reported to be signi
cant, even under
experimental conditions, both with image-based [
17
] and with surface-based
algorithms [
29
]. Reported success rates are below 90 %, and clinical cases would
