Biomedical Engineering Reference
In-Depth Information
other classifying quality) represented in each voxel and using that information to
assign color and other visual properties to the voxel.
In surface rendering, only part of the 3D dataset is utilized for the reconstruc-
tion of an image. The data to reconstruct the image is selected by defining a thresh-
old. Typically, the voxels with a value below the threshold are discarded from the
data, and the voxels with a value equal to or above the threshold are selected for
the rendering. The main advantage of surface rendering is its speed, since a com-
paratively small amount of computational power is needed to generate images in
a reasonable amount of time. However, the thresholding technique is based on the
assumption that each volume element represents only one type of tissue. Hence, it
incorrectly classifies voxels that represent mixed tissue interfaces. The thresholding
technique is also susceptible to noise introduced during scanning. A small amount
of noise can modify attenuation values and create the appearance of soft tissue that
actually represents mostly bone.
Unlike surface rendering, volume rendering does not make use of a surface
representation but uses a technology originally developed for motion picture com-
puter animation. When a volume rendering algorithm is used, certain properties
are assigned to each voxel on the basis of its value. In CT, it is the average of all
the attenuation values contained within the corresponding voxel. This number is
compared to the attenuation value of water and displayed on a scale of arbitrary
units named hounsfield units (HU) named after British electrical engineer Godfrey
N. Hounsfield. When properties are assigned a certain value, volume rendering
uses the histogram of these values. In a typical CT histogram, the
x
-axis represents
the possible voxel values in the data, and the
y
-axis, the number of voxels with
that specific value. Certain values can be related to specific tissue compositions
when the properties of a voxel value are determined. Each of these tissue composi-
tions has specific properties. Partial rendering enables mixing two different tissue
compositions (totaling 100%) to establish the properties of a border voxel. Thus, a
voxel can partially belong to the surface of interest and can have properties based
on the percentages of the properties of the two tissues involved.
8.4.4 ImageQuality
Quality of the image produced during the imaging process limits visibility of ana-
tomical details and other small objects in the body. The level of detail in an image is
related to the size of objects. The general range of details for each imaging modality
is determined by the limitations of the technology, design characteristics, and the
selected operating factors. Problems associated with digitizing images could com-
promise the quality of the image. For example, some of the weaknesses of CCDs
that could result in decreased image quality are:
1.
Fading
: Although the coupling process is quite effi cient, moving the charges
along a row of many pixels adds up to a noticeable loss of charge.
2.
Blooming
: If too many photons strike a CCD element, it gets saturated and
some of the charge leaks to adjacent pixels.
3.
Smearing
: If light strikes the sensor while a transfer is taking place, it can
cause some data loss and leave streaks behind bright areas of the image.
