Computed Tomography 2
Cone Beam CT
Cone beam CT uses a large fan beam (>20 degrees) and a large array of detectors, meaning a large amount of the x-axis can be included in one scan and the table does not need to be moved. For example, a recent model uses 64 detectors, each of 0.5cm meaning 32cm of the z axis can be imaged. CBCT is used in perfusion studies of the head, kidneys and heart, producing images of high temporal resolution. CBCT is also very common in radiotherapy.
However, using cone beam CT introduces increased scatter and reduced image resolution.
Cardiac CT
Imaging the beating heart has proven difficult, because such a small window is required to image the heart without motion artefact. This window would ideally be around 100ms. Although electron beam CT comes close with a window of 50ms, fifth generation CT is expensive. Thus, third generation CT is used but by employing prospective cardiac gating. Prospective cardiac gating uses an ECG linked to the x-ray tube, which activates the x-ray tube during the most inactive part of the cardiac cycle, i.e. end of diastole.
Cardiac CT is used in assessing coronary artery stenosis and other abnormalities, however an image of the full cardiac cycle is still not achieved. Artefacts result using prospective cardiac gating may result from irregular heart beating, although beta blockers can be used to correct this.
CT Perfusion Imaging
CT perfusion imaging provides high temporal resolution images that can be used to demonstrate vascular perfusion and other parameters related to the blood flow to an organ. The CT scanner acquires images in real time repeatedly to quantify the flow of iodine-based contrast through the organ. A series of images is taken without contrast to create a parametric map, and a second series is taken with contrast. The images are then reconstructed into four types of images; cerebral blood flow, cerebral blood volume, mean transit time and Tmax. The scan, however, is continuous, meaning CT perfusion imaging is a high dose procedure.
CT Angiography
Conventional fluoroscopy-based angiography has overtime begun to be replaced by computed tomography angiography, which provides higher contrast resolution. CT angiography requires only intravenous access, as opposed to intra-arterial DSA which is more invasive.
Reconstruction Methods
In CT, there are three common reconstruction methods after the voxel number has been assigned.
Filtered Back Projection
Back projection is one type of reconstruction. The acquired data is 'back projected' after the scan onto a matrix. The data is pushed back along the path it took in space. The brightness is summed and all pixels are added up and normalised. The more attenuated an x-ray is, the brighter the stripe of back projection. Back projection produces a 'star effect' which is removed using a convolution filter, giving the reconstruction method filtered back projection.
Iterative Reconstruction
Iterative reconstruction is the second type of reconstruction which involves a series of iterations on a matrix. It begins with a 'guess' image which is compared with the actual image. By going through a series of interactions, the guess image becomes more and more alike the actual image, continuing until a final, accurate depiction of the actual image.
Different CT manufacturers use different components (x-ray tubes, detectors, etc.) within their CT machines. This means if we use different scanners to get similar image quality for a particular scan, dose will vary. This used to be a problem, because there was no regulation standard. So, to rectify the situation, professional bodies introduced the term CTDI which is useful in comparing different machines, protocols and their radiation output.
CTDI can be measured in an ionisation chamber inserted into a standard phantom. Initially we had CTDI100 which was a measurement using a 100mm pencil ionisation chamber. However, this had several limitations, one being that it didn't represent the geometry of a patient.
Then a new index known as weighted CTDI was introduced, where two measurements were taken at two locations within the chamber and a weighted average was calculated. This better represented the patient dose profile, but was only useful for axial CT, not helical.
Current, we use two quantities: CTDIvol and DLP.
CTDIvol was introduced to take into account the pitch of a helical CT scanner. As CTDIvol is basically just weighted CTDI with the inclusion of pitch, CTDIvol for an axial scanner is equal to weighted CTDI. CTDIvol is a standardised parameter to measure the radiation output of a CT scanner, but is not a measure of patient dose. Pitch is inversely proportional to dose. Thus, CTDIvol is calculated by dividing CTDIW by pitch. It is measured in mGy.
However, CTDIvol does not take into account the scan length. This means for a 10cm and 40cm length, the CTDIvol will be the same, even though dose for a 40cm scan would be much higher.
So, DLP was introduced. DLP stands for dose length product and is the product of CTDIvol by the length of the scan. It is measured in mGy/cm. DLP is nearly proportional to dose and can be used to estimate dose using k factors, measured in mSv/mGy-cm.
Expected CTDIvol and DLP are displayed before the scan, as calculated from patient thickness information from the CT radiograph. ARPANSA provides some recommended values, e.g 400mSv for a head CT. The acquired value is then compared with these predicted values. The dose notification level is an alert given when a higher than normal value is provided and gives the opportunity to make changes, which can be overridden if necessary. The dose alert level is an alert which requires specific action by the operator before continuing.
CT Artefacts
Streak artefact is when a high density material has attenuation levels which exceed the dynamic range of the detector system, e.g. a metal tooth filling. To overcome streak artefact, kVp should be increased and shorter exposure time should be used.
Partial volume is when the detector element is large enough to encompass more than one type of tissue, e.g. bone and soft tissue, and the resultant voxel grey scale value is proportional to the average beam attenuation of these tissues. Smaller voxel size, causing higher spatial resolution, prevents partial volume artefact.
Ring artefact is when there is a dead or miscalibrated detector, which appears as a ring on multiple slices in the same location.
Differences between diagnostic and therapy CT
CT, particularly cone beam CT, is also applicable in radiotherapy.
Cone beam CT uses a large fan beam (>20 degrees) and a large array of detectors, meaning a large amount of the x-axis can be included in one scan and the table does not need to be moved. For example, a recent model uses 64 detectors, each of 0.5cm meaning 32cm of the z axis can be imaged. CBCT is used in perfusion studies of the head, kidneys and heart, producing images of high temporal resolution. CBCT is also very common in radiotherapy.
However, using cone beam CT introduces increased scatter and reduced image resolution.
Cardiac CT
Imaging the beating heart has proven difficult, because such a small window is required to image the heart without motion artefact. This window would ideally be around 100ms. Although electron beam CT comes close with a window of 50ms, fifth generation CT is expensive. Thus, third generation CT is used but by employing prospective cardiac gating. Prospective cardiac gating uses an ECG linked to the x-ray tube, which activates the x-ray tube during the most inactive part of the cardiac cycle, i.e. end of diastole.
Cardiac CT is used in assessing coronary artery stenosis and other abnormalities, however an image of the full cardiac cycle is still not achieved. Artefacts result using prospective cardiac gating may result from irregular heart beating, although beta blockers can be used to correct this.
CT Perfusion Imaging
CT perfusion imaging provides high temporal resolution images that can be used to demonstrate vascular perfusion and other parameters related to the blood flow to an organ. The CT scanner acquires images in real time repeatedly to quantify the flow of iodine-based contrast through the organ. A series of images is taken without contrast to create a parametric map, and a second series is taken with contrast. The images are then reconstructed into four types of images; cerebral blood flow, cerebral blood volume, mean transit time and Tmax. The scan, however, is continuous, meaning CT perfusion imaging is a high dose procedure.
CT Angiography
Conventional fluoroscopy-based angiography has overtime begun to be replaced by computed tomography angiography, which provides higher contrast resolution. CT angiography requires only intravenous access, as opposed to intra-arterial DSA which is more invasive.
Reconstruction Methods
In CT, there are three common reconstruction methods after the voxel number has been assigned.
Filtered Back Projection
Back projection is one type of reconstruction. The acquired data is 'back projected' after the scan onto a matrix. The data is pushed back along the path it took in space. The brightness is summed and all pixels are added up and normalised. The more attenuated an x-ray is, the brighter the stripe of back projection. Back projection produces a 'star effect' which is removed using a convolution filter, giving the reconstruction method filtered back projection.
- Advantage: quick, less powerful computers are required
- Disadvantage: image quality is poor and images are noisy
Iterative Reconstruction
Iterative reconstruction is the second type of reconstruction which involves a series of iterations on a matrix. It begins with a 'guess' image which is compared with the actual image. By going through a series of interactions, the guess image becomes more and more alike the actual image, continuing until a final, accurate depiction of the actual image.
- Advantage: high resolution, low noise, fewer artefacts
- Disadvantage: requires more computing power
Cone Beam Reconstruction
Cone beam reconstruction utilises filtered back projection, but encompasses not only the fan angle by keeps track of the beam divergence in the z direction, i.e. the cone angle divergence. Back projection is undergone, where the acquired data is back projected onto a matrix. The data is pushed back along the path it took in space and the brightness is summed and normalised. The brighter the back projection stripe, the greater the x-ray attenuation. In cone beam reconstruction, back projection occurs not only along the fan angle, but also the cone beam angle. Cone beam reconstruction reconstructs the entire volume data set simultaneously. Cone beam reconstruction can result in cone beam artefact.
Isotropic Reconstruction
Isotropic reconstruction enables the reconstruction of images into any plane. Using multi detector CT, thin slices can be acquired which have isotropic spatial resolution. This means the slices can be stacked on top of one another and reconstructed into volume images in another plane, e.g. sagittal. 3D imaging is highly useful in clinical diagnosis.
Dose Modulation in CT
In CT, kVp is generally held constant and mA is changed. Dose is significantly higher in CT and thus dose must be minimised in order to reduce patient radiation exposure. Automatic exposure control is used to modulate mA based on single slice and also the z axis of the scanner. The aim of dose modulation is to produce images of high quality at the lowest possible dose. A disadvantage of lower mAs is that the images become noisy and contrast resolution is lost. However, despite this, dose modulation is almost always used. Dose modulation can be set with help from the CT radiograph.
CTDI
Different CT manufacturers use different components (x-ray tubes, detectors, etc.) within their CT machines. This means if we use different scanners to get similar image quality for a particular scan, dose will vary. This used to be a problem, because there was no regulation standard. So, to rectify the situation, professional bodies introduced the term CTDI which is useful in comparing different machines, protocols and their radiation output.
CTDI can be measured in an ionisation chamber inserted into a standard phantom. Initially we had CTDI100 which was a measurement using a 100mm pencil ionisation chamber. However, this had several limitations, one being that it didn't represent the geometry of a patient.
Then a new index known as weighted CTDI was introduced, where two measurements were taken at two locations within the chamber and a weighted average was calculated. This better represented the patient dose profile, but was only useful for axial CT, not helical.
Current, we use two quantities: CTDIvol and DLP.
CTDIvol was introduced to take into account the pitch of a helical CT scanner. As CTDIvol is basically just weighted CTDI with the inclusion of pitch, CTDIvol for an axial scanner is equal to weighted CTDI. CTDIvol is a standardised parameter to measure the radiation output of a CT scanner, but is not a measure of patient dose. Pitch is inversely proportional to dose. Thus, CTDIvol is calculated by dividing CTDIW by pitch. It is measured in mGy.
However, CTDIvol does not take into account the scan length. This means for a 10cm and 40cm length, the CTDIvol will be the same, even though dose for a 40cm scan would be much higher.
So, DLP was introduced. DLP stands for dose length product and is the product of CTDIvol by the length of the scan. It is measured in mGy/cm. DLP is nearly proportional to dose and can be used to estimate dose using k factors, measured in mSv/mGy-cm.
Expected CTDIvol and DLP are displayed before the scan, as calculated from patient thickness information from the CT radiograph. ARPANSA provides some recommended values, e.g 400mSv for a head CT. The acquired value is then compared with these predicted values. The dose notification level is an alert given when a higher than normal value is provided and gives the opportunity to make changes, which can be overridden if necessary. The dose alert level is an alert which requires specific action by the operator before continuing.
CT Artefacts
- Beam Hardening
- Streak Artefact
- Partial Volume
- Ring Artefact
Streak artefact is when a high density material has attenuation levels which exceed the dynamic range of the detector system, e.g. a metal tooth filling. To overcome streak artefact, kVp should be increased and shorter exposure time should be used.
Partial volume is when the detector element is large enough to encompass more than one type of tissue, e.g. bone and soft tissue, and the resultant voxel grey scale value is proportional to the average beam attenuation of these tissues. Smaller voxel size, causing higher spatial resolution, prevents partial volume artefact.
Ring artefact is when there is a dead or miscalibrated detector, which appears as a ring on multiple slices in the same location.
Differences between diagnostic and therapy CT
CT, particularly cone beam CT, is also applicable in radiotherapy.
- The gantry bore must be larger to accomodate for immobilisation devices
- The table must be flat to match the flat couch of radiotherapy
- An accurate CT number must be calibrated as physical or electron density can be then calculated, and density information is important in therapy treatment planning
- Accurate laser positioning must be available in the CT machine for treatment set up reproducibility
- Artefact suppression is important
Spatial Resolution
Spatial resolution is the smallest object which can be seen at high contrast in the CT. It is defined in line pairs visible or Modulation Transfer Function. It depends on:
- Pixel size
- Focal spot size
- Size of detector
- Slice thickness
Contrast Resolution
Contrast resolution is a measure of the lowest contrast object visible for a given size.
Uniformity
Uniformity measures the average HU in a number of regions of interest in an image. It should be as similar as possible. Uniformity allows us to check the CT number of that of water.
Linearity
Linearity ensures that reconstructed pixels are the correct size by measuring a known length.
HU Accuracy
HU accuracy checks the accuracy of the CT number. It is undergone by measuring the HU of known objects, which should be within 40 of the expected value.
Comments
Post a Comment