R_Bibb_Medical_Modelling_The_Application_of_Adv.pdf

Medical imaging for rapid prototyping 11
performed much more rapidly and, consequently, three-dimensional CT
scans are frequently referred to as helical CT. In addition, modern CT
scanners employ multiple arrays to enhance the rate of data capture and
improve three-dimensional volume acquisition.
CT images are generated as a grey scale pixel image, just like a bitmap
computer image. If the distance between a series of axial images, called the
slice thickness, is known, they can be interpolated from one image to the
next to form cuboids, known as voxels. Therefore, a three-dimensional CT
scan generates a voxel representation of the human body. Software can be
used to re-slice these voxel data sets in axes perpendicular to the long axis
enabling different cross-sectional images to be generated from the original
axial data. This is typically done in the sagittal and coronal planes; however,
images may be generated in any plane.
The radiographers who conduct CT scans have specifi c parameters and
settings for different types of scan. These are standardised and referred to
as protocols. When embarking on using CT data for medical modelling it
may be helpful to discuss it fi rst with the radiographers and they may well
develop a protocol specifi cally for medical modelling. CT scans are time
consuming, expensive and potentially harmful so every care must be taken
to ensure that the scan is conducted correctly the fi rst and only time.
2.2.2 Partial pixel effect
When CT data is captured, the resulting images are divided up into a large
number of pixels (typically a 512 × 512 matrix). Each pixel is a shade of
grey that relates to the density of the tissue at that location. The resulting
images are, therefore, an approximation of the original tissue shapes according
to their density. The quality of that approximation is a function of the
number and relative size of the pixels as well as other aspects of the CT
scanner. The discrete size of the pixels means that edges between different
anatomical structures are to some degree affected by this image quality.
The effect can be to ‘blur the edges’ due to the partial pixel effect.
If the boundary between two different structures crosses a given pixel,
that single pixel cannot represent both densities. Instead that pixel displays
an intermediate density which is somewhere between the two. The effect
can be illustrated by considering the shape in Fig. 2.2, which consists of
two densities, low being grey and high being white. When the CT scan is
performed, the cross section is broken down into pixels as shown in
Fig. 2.3. In this view, it can be seen that some squares contain both high-
and low-density areas. These pixels will, therefore, be shown as an intermediate
grey depending on their relative proportions. This leads to the
partial pixel effect that can be seen in the tomographic image that results,
shown in Fig. 2.4.