Biomedical Engineering Reference
In-Depth Information
fetus has a three-dimensional (3D) structure, 2D ultrasound can depict only
sectional images. This means that 2D ultrasound cannot depict the eyes, nose,
and mouth on an image (Fig. 2.8), for example. An examiner has to imagine
the 3D structure of the fetus in his or her mind for an accurate diagnosis. Ho-
wever, it is not easy for most examiners to imagine the 3D structure from 2D
ultrasound images, and many abnormalities are overlooked in daily practice.
Fetal growth is evaluated by estimating fetal body weight. However, the esti-
mation is sometimes inaccurate because it is based on only several lengths,
biparietal diameter (BPD), antero-posterior trunk diameter (APTT), trans-
verse trunk diameter (TTD) (Fig. 2.10), and so on.
2.2.2
The Development of Three-Dimensional Ultrasound
Three-dimensional ultrasound for a fetus was introduced in the 1980s [10-12]
to overcome the limitations of conventional 2D ultrasound. Many technologies
for 3D untrasound were developed in the 1990s. [13,14]. The following 3D
ultrasound imaging methods are available at present:
1. Computer processing
Section reconstruction
Surface rendering
Volume rendering
2. Defocusing lens method
3. Real-time ultrasonic beam tracing
Computer Processing
Computer processing is the oldest and the most popular method for 3D ul-
trasound. Three steps (3D data acquisition, 3D data set construction, and
projection/display) are required in this method. Three-dimensional data are
usually acquired through movement of a probe. A 3D probe contains a con-
vex probe of conventional 2D ultrasound, and the convex probe swings in
the 3D probe to acquire 3D data. A 3D data set is then constructed in the
computer. In 3D ultrasound by computer processing, many kinds of images
can be generated from the 3D data set. Surface rendering was initially in-
troduced for a 3D fetal image in the 1980s [10-12]. Good 3D images can be
obtained in X-ray CT and MRI by surface rendering, because a boundary
between different tissues can be clearly identified. But in ultrasonography,
the boundary is often obscure and 3D images by surface rendering are often
insucient. Volume rendering usually makes a better 3D image (Fig. 2.11)
and is used more often nowadays than surface rendering. Not only a surface
image but also a skeletal image (Fig. 2.12) can be obtained by both surface
rendering and volume rendering by selecting parameters in their processes.
Any arbitrary section can be obtained by section reconstruction from the 3D
data set. In practice, three-orthogonal sectional images are usually displayed
simultaneously (Fig. 2.13).
