Page 421 - Biomedical Engineering and Design Handbook Volume 2, Applications

COMPUTER-INTEGRATED SURGERY AND MEDICAL ROBOTICS 399

further use unless scanned. CT and MRI images are used to visualize anatomical structures, with CT
best suited for bony structures and MRI best suited for soft tissue. They consist of a series of two-
dimensional parallel cross-sectional images with high spatial resolution, little geometric distortion
and intensity bias, good signal-to-noise ratio, and a wide field of view. Typical data sets consist of
80 to 150 images of size 512 × 512 12-bit gray-level pixel images with pixel size of 0.4 × 0.4 mm
at 1- to 2-mm intervals. They can be used to visualize anatomical structures, perform spatial mea-
surements, and extract three-dimensional anatomical models. NMT images show functional
anatomy, such as nerve activity, and are mostly used in the brain. They also consist of a series of two-
dimensional parallel slices, although their quality is lower. They are usually viewed in conjunction
with MRI images. The main drawback of preoperative images is that they are static and don’t always
reflect the position and orientation of anatomical structures which have moved between the time the
images were taken and the surgery is performed.
Intraoperative imaging devices include fluoroscopic x-ray, ultrasound, and video image streams
from endoscopes, laparoscopes, and surgical microscopes. Fluoroscopic x-ray is widely used in
orthopedics to visualize and adjust the position of surgical instruments with respect to bones, or to
locate kidney stones. The images are obtained from a mobile C-arm unit, which allows capturing
two-dimensional projection images from different viewpoints while the patient lies on the table. The
circular images are usually displayed on a video monitor. They have a narrow field of view (6 to 12 in,
400 pixels in diameter), limited spatial resolution and contrast, and present varying, position-depen-
dent intensity and geometric distortions. They are mostly used for qualitative evaluation, and have
cumulative radiation as a side effect. Ultrasound images (both static and as sequences) are used to
obtain images of anatomy close to the skin. Unlike x-ray images, they have no ionizing radiation, but
present significant imaging artifacts, such as speckling, noise, and spatial distortion. They also have
a narrow field of view and have the resolution and image quality of the standard video monitor
where they are displayed. Video image streams became commonplace with the introduction of min-
imally invasive surgery in the 1980s. They are used to support tumor biopsies, gall bladder removals,
and colon explorations, among many others. They allow the surgeon to visualize in real time
anatomy and surgical instruments inserted in a body cavity. The limitations of these imaging
devices are that they have a narrow field of view (about 3 in), have no depth perception, uneven illu-
mination, distortion due to the use of wide-angle lenses, and require direct line of sight. Surgeons
must learn how to move and point the camera while respecting various point-of-entry and location
constraints. The main advantage of intraoperative images is that they provide an up-to-date image
of the surgical situation. However, the field of view and image quality are far inferior to preopera-
tive images. More recent intraoperative imaging devices include surgical open MRI, surgical CT,
and three-dimensional (3D) ultrasound, which overcome some of the limitations of the more com-
mon imaging devices.
The main limitation of current practice is that there is no quantitative correlation between high-
quality preoperative images and intraoperative images. The surgeon must mentally establish the spa-
tial correlation between the images and base decisions on this correlation.

14.3.2 Image Processing, Visualization, and Modeling

After image acquisition, the first task is usually visualization for diagnosis, evaluation, and planning.
The visualization can take place on displays other than those of the devices where they were
acquired, and can require various image-processing techniques for better evaluation. These include
image balancing and enhancement, distortion and contrast correction, denoising, and spatial aggre-
gation. For example, individual two-dimensional x-ray and ultrasound images can be processed
using an array of standard image processing techniques to improve their clinical value. They can be
visualized using zooming, cropping, and other imaging techniques. They can also be combined to
create new, multimodal images.
Visualization of CT, MRI, and nuclear medicine images can greatly benefit from specialized visu-
alization techniques, since they are series of two-dimensional cross sections. Instead of having the