Geometric Model Reconstruction - Computational Hemodynamics: Theory, Modelling and Applications - page 43

issues relevant to each modality and the anatomical regions for which segmentation
methods can be applied. New techniques are continuously emerging through rapid
advances in technology and these techniques are covered in peer reviewed journals
such as Current Cardiovascular Imaging Reports, IEEE Transactions on Medical
Imaging , Medical Image Analysis , that are dedicated to medical imaging.
MRI scans use magnetic fields, and radio waves to obtain cross-sectional im-
ages of the body. During a scan an electric current passes through coils of wires to
generate a magnetic field. Hydrogen protons of water molecules inside the body
that normally spin in random directions are then aligned with the magnetic field.
A short burst of tuned radio waves is sent through the body which momentarily
changes the quantum state of the hydrogen protons (e.g. flips the spin of the proton).
When the radio wave stops, the proton returns to its original orientation and in do-
ing so echo's its own radio signal that a scanner detects and deciphers into images.
This means that different tissue structures produce different pulse sequences, lead-
ing to contrast changes for a number of tissue parameters. In addition, anatomical
and physiological variation between subjects requires different pulse sequences to
achieve the correct contrast.
CT scans use multiple x-rays taken at thin cross-sections in the region of interest
along the person's body forming slices (like slicing a loaf of bread). During a scan
x-ray beams consisting of photons are absorbed or redirected (i.e. scattered) by ma-
terial in the body which reduces the strength of the x-ray beam. Electronic detectors
collect the x-ray information from each cross-section and send them to a computer
that combines them into a single image. CT scans produce images with resolutions
equal to or better than MRI. However soft tissue contrast in CT is not as good as in
MRI, but is superior for imaging bone and bone tumors. Since CT scanners use x-
rays which are a form of ionising radiation, its cumulative use has associated risks
that are unavoidable and therefore CT scans are only performed where the benefit of
the examination outweighs any potential risks. Both MRI and CT are non-invasive
techniques however a contrast dye is sometimes injected into the body via one of
the veins during the scan, referred to as an MR- or CT- angiograph. The contrast dye
highlights the circulatory pathway and allows detection of the coronary arteries on
the x-ray images. This type of scanning is referred to as angiography .
The scanned images produced are in a variety of formats but the most generic
and common format is DICOM which contain both image data and patient infor-
mation. These images are a series of stacked 2D pixels separated by a slice thick-
ness, and when combined provide 3D volumetric data. The images contain pixels of
greyscale values with respect to different organs or tissues. These values represent a
mapping (Fig. 3.1 ) of the linear x-ray attenuation coefficient in CT or a measure of
the radio-density in MRI. In medical imaging a Hounsfield scale with units of HU
is used, while in image processing the greyscale is numbered between 0 and 255.
In a complete MRI/CT scan a set of 2D cross-sectional images, each separat-
ed by a thickness is produced. When the set of images are collated together, 3D
volumetric information can be obtained. Each 2D image is made up of an array of
pixels—the smallest element of the image. When the slice thickness between two
images is considered, these pixels become voxels ( volume of pixels Fig. 3.2 ). The