Topic 1: Slice Thickness & Interval
.png)
Slice Thickness
Slice thickness can be broken into two categories:
-
Detector Slice Thickness: This is the z-axis thickness of each individual detector slice within the detector array. The smaller the slice thickness, the better the spatial resolution of the images. Most modern scanners have a slice thickness <1mm.
-
Reconstructed Slice Thickness: This is the depth of the voxels on a multiplanar reconstructed image. The operator can select any thickness equal to or higher than the detector slice thickness. Each slice makes up an image. The slices are then stacked together to form a series of images which the viewer can scroll through. An analogy for this is cutting a loaf of bread into slices of a certain thickness, then stacking them back together in a loaf and viewing each slice of bread one by one.
Slice Interval
This is the distance between the centre of two adjacent slices within a series of images.
For a contiguous image, the slice interval will match the slice thickness, so the next slice will start where the previous slice ends.
If the slice interval is less than the slice thickness, then this will result in overlapped slices. This will create more images in the series.
If the slice interval is more than the slice thickness, then this will result in non-contiguous slices (gaps between the slices). This will create less images in the series, however the anatomy that falls within the gap will not be visible.
.jpg)
The majority of multiplanar reconstructions used in CT are contiguous, because it is the optimal balance between not having too many images (as we get with overlapped slices) and not leaving any anatomy out of the images (as we get with non-contiguous slices).
Overlapped slices can be useful in some instances, especially when viewing very small and intricate anatomy such as small blood vessels on an angiogram. Non-contiguous slices run the risk of missing anatomy and pathology, so they are rarely used
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 2: SFOV & DFOV
Scan Field of View (SFOV) is the diameter in x/y-axis around the isocentre where data is captured during a scan. This is set in the scanning protocol as either small (approx 25-32cm) or large (approx 50cm).
Display Field of View (DFOV) is the diameter of a reconstructed series. It determines how much anatomy is displayed in the x/y-axis, and where the image is centred.
Figure 1a: An axial slice showing the SFOC (red) and DFOV (blue)
Figure 1b: Images reconstructed from the same raw data with a large 50cm DFOV (left) and small 20cm DFOV (right)
Where SFOV is embedded in the raw data and cannot be changed, DFOV can be altered for each reconstructed series. This is important to be aware of, because if you find that the DFOV on your initial reconstructed series is too small and has clipped off an important part of anatomy, you can reconstruct another series using the same raw data with a larger DFOV to include the area of interest, provided it falls within the SFOV. The viewer can also change the DFOV on any image in a DICOM viewer using a magnification tool.
Along with slice thickness, DFOV has a significant impact on the spatial resolution of an image. The larger the DFOV, the lower the spatial resolution. This is because there are a fixed number of pixels in a reconstructed image, so when the DFOV is increased, the pixel size must also increase. Therefore it is important to keep the DFOV as small as possible, while still including all the relevant anatomy.
“Zooming” an image in a viewport will not change the spatial resolution - the original DFOV needs to be changed in the reconstructed image from the raw data. An analogy would be the scale of a map:
If we increase the scale of a map, we are seeing a larger area of land per mm on the map. Larger objects may be visible, such as main highways, suburbs and cities. If we decrease the scale, we are seeing a smaller area of land per mm on the map and we are able to visualise more intricate details, such as residential streets and alleyways and building numbers.
However, if we were to place a magnifying glass over the large scale map, we may be able to see a small section of the area, but we won’t be able to visualise any of the smaller details that we see on the map with a decreased scale. See images below.
Figure 2a: A large scale map (left) and a zoomed-up section of the map (right). This is similar to having a large DFOV image and zooming it up in the viewport.
Figure 2b: A small scale map, representing the same area as the zoomed-up section in figure 2a. This is similar to reconstruction a series with a small DFOV.
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 3: kVp & mA
kVp determines the average energy of the x-ray beam. This determines how much attenuation will occur within the body from different anatomical structures. A high kVp beam will have less overall attenuation compared to a low kVp beam, however the differences are not linear. Anatomical structures with a higher atomic number will have a proportionally higher level of attenuation at low energies (this is described by their attenuation coefficient). This results in higher contrast resolution at low energies, because the differences in attenuation between two structures may be proportionally greater (see figure below).
Figure 1: Note the attenuation through liver as kVp increases from 100 to 140 only increases by 1 point, whereas for Iodine it increases by 2 points, creating less contrast between the two structures at 140kVp.
As always in CT, there is a trade-off. While low kVp produces images with high image contrast, they also have more image noise. This is because more photons are being attenuated by the patient and not reaching the detector, resulting in a lower signal. Beam hardening artefact is also greater with low kVp scans, because there are less high energy photons to penetrate dense materials.
Because of this trade-off, the operator may favour a low kVp for certain scans, but a high kVp for others. As a general rule of thumb, you want to aim for lower kVp when using IV contrast to improve contrast resolution, and higher kVp for non-IV contrast scans and/or dense body areas to reduce noise and beam hardening.
mA is the amount of current applied across the x-ray tube, and directly affects the amount of photons produced. This determines the number of photons that will be absorbed by the patient and reach the detector, and therefore changing the mA will change the image noise and patient dose.
The relationship between mA and dose is linear; if we double the mA (while keeping all other factors the same), the dose will double.
The relationship between mA and noise can be expressed by:
mA = 1/√Noise
If we double the mA, we decrease noise by a factor of 0.7.
There is a common misconception that a higher kVp will result in higher CTDIvol, however this is not usually the case. It is true that if all other factors remain constant and kVp is increased, then dose will increase. However, all modern CT scanners today use mA/dose modulation, which automatically adjusts the mA to achieve a desired noise level if another scanning factor is changed, such as rotation speed, pitch and kVp. As long as the patient size allows the mA to be adjusted within the minimum and maximum settings, changing the kVp should result in only a minor change to DLP and CTDIvol. For cases of very small or large patients, if the kVp has not been appropriately set, then the dose may be too low or too high as the mA modulation will not have the capacity to work efficiently.
It is therefore important to check the mA modulation table for each scan to ensure the most appropriate kVp has been selected.
Figure 2: mA modulation table showing the mA "maxing out" (left) due to the kVp setting being too low, and a more appropriate mA modulation (right).
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 4: Isocentre
The isocentre is the central focal point of the CT gantry, as depicted in the figure below. It is where the patient positioning lights intersect in the x and y-axis.
Figure 1: Depiction of Isocentre relative to the CT Gantry and SFOV (below), and the relative photon flux within the SFOV (above).
The photon flux (i.e. concentration of photons) is highest at the isocentre, and gradually decreases as we move towards the edge of the SFOV. This is due to a bowtie-shaped filter that attenuates photons at a higher proportion near the edge of the SFOV before passing through the patient (see figure below). This is a dose-saving measure; because the majority of anatomical objects we scan are circular, the peripheral edges of the objects are less thick and require less photons to generate an image with a similar noise level.
.jpg)
Figure 2: A bowtie filter in a CT system. Note the higher level of attenuation towards the edge of the filter, correlating to the edge of the SFOV.
The bowtie filter, however, can backfire if not used correctly. If the patient is not positioned in the isocentre of the gantry, the photon flux will be disproportionate relative to the body thickness. This will result in two negative outcomes:
-
There will be an over-exposure of the peripheral body region that is closer to the isocentre, resulting in more absorbed dose.
-
There will be an under-exposure of the central body region that is further away from the isocentre, resulting in higher image noise.
Figure 3: Correct patient positioning (left) vs. incorrect patient positioning (right). Note the overdosing of the edge of the object within the isocentre.
The most common trap radiographers fall into in this respect is not setting the table height correctly, so the patient is outside the isocentre along the y-axis. X-axis (left-right) positioning is generally not as variable due to the fixed location of the table in this plane.
The biggest impact on image quality and dose you can have as a radiographer is to ensure your patient is positioned in the isocentre. So pay extra attention to this area of your job. As a general rule of thumb, raising the table height so the isocentre is at the patient’s mid-axillary line will result in the most uniform dose distribution across the body. Doing this will also improve the likelihood of including the patient’s entire anatomy in the SFOV.
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 5: Image Quality Measurements
Image Noise
Noise can arise from many different sources during the image acquisition process, and is caused by interference during the sending and receiving of data. The level of noise in an image is inversely proportional to the signal, as described by the signal-to-noise ratio (SNR). As we increase the signal (i.e. the tube output) of a scan, the noise level decreases.
If we were to acquire a CT image with zero noise of a completely homogeneous object (for example a water phantom), the Hounsfield Unit of each pixel would be identical. That is, there would be no deviation from what we see in the image compared to the object being imaged. Unfortunately in the real world, capturing an image with zero noise is impossible.
In medical imaging, noise is measured in standard deviations (𝜎). The higher the standard deviation, the greater the discrepancy of each pixel HU from the original object. When viewing an image with a high image noise level, the appearance is typically “grainy”. As noise decreases, the image becomes more “smooth” and homogeneous.
Figure 1: Axial images of a water phantom showing different noise levels: (Left) High noise, (Middle) Average noise, (Right) Low noise
Contrast Resolution
The ability to distinguish between two adjacent objects in an image is a measure of the contrast resolution. The greater the difference in HU between the two objects, the better the contrast resolution.
Different anatomical structures will have a natural level of contrast depending on their density. For example, air-filled lungs (black) are easily distinguished from the patient’s ribs (white). However distinguishing between soft-tissue and a blood vessel is much more difficult because of their similar HU. The specific term for this is low contrast detectability, and improving this is a constant goal for CT vendors.
IV contrast is often used to improve the contrast resolution between blood-filled vessels and organs and the surrounding soft-tissue and organ parenchyma.
The beam energy (kVp) is inversely proportional to image contrast - a low energy beam results in a higher level of contrast (see mA & kVp topic for more details).
Figure 2: An axial image with low image contrast (left), compared to high image (right) in the pulmonary trunk
Spatial Resolution
The ability to visualise small objects in an image is a measure of the spatial resolution. The most common way to measure spatial resolution is in line pairs per mm (LP/mm). The actual number provided by vendors (e.g. 28 LP/mm) is a theoretical number and not accurate to what can be actually visualised in practice. It is useful, however, in describing how close two parallel lines can be next to each other and still be clearly visualised.
To test this, a CT phantom is scanned and images are reconstructed perpendicular to the plane of the line pairs. As the line pairs are spaced closer together, it becomes increasingly difficult to distinguish individual lines. An image with high spatial resolution would be able to clearly display line pairs that are spaced closer together.
The term pixel size or pixel number is often interchangeable with spatial resolution. The larger the pixel size, the lower the spatial resolution. This is because there is a larger distance between each pixel, and therefore any small changes in the object are not accurately displayed.
Depending on your scanner, it is often not possible to change the pixel number of your image - this is usually locked at 512x512. What you can change, however, is the area of anatomy displayed within the 512x512 matrix; this is determined by the DFOV (see the earlier topic on SFOV and DFOV for more information on this).
Figure 3: An axial image of a phantom showing line pairs
Temporal Resolution
Temporal resolution describes the ability to image a moving object within a set timeframe. The faster an image is acquired, the better the temporal resolution. If temporal resolution is low, a moving object may experience more significant motion artefacts.
This is an important factor to consider when imaging structures that have involuntary movements, such as the heart or bowel wall.
Figure 4: A CTCA with low Temporal Resolution (top - blue) and high Temporal Resolution (bottom - red)
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 6: Iterative Recontruction
Iterative Reconstruction (IR) has become standard on all CT scanners in the past decade. Put simply, IR reconstructs the raw data into a DICOM image in a much more efficient way compared to the traditional Filtered Back Projection (FBP) method. This results in images with much less noise, which means we can use IR to either:
a) Scan at the same dose and obtain images with less noise compared to FBP; or
b) Scan at a lower dose and obtain images with a similar noise compared to FBP
Most practices opt for the second option, which has resulted in a dramatic reduction in CT dose levels over the past decade.
Because IR is a post-processing tool, we can retrospectively apply different levels to a scan to generate images with different noise levels. Each vendor has different options for IR levels; some only have a few (low, medium, high) while others have a range from 0-100% with 10% increments. When we select an IR level, we are telling the scanner to reconstruct the image with a specific blend of IR and FBP.
For example, an image with 60% IR will use FBP for the remaining 40%. This blend allows us to create images that still have the traditional FBP appearance, but at the same time reduce noise in the image. A highly-weighted IR image will reduce a lot of noise, but starts to degrade in image quality in other areas. The most common description of a high IR image is that it has a “plastic” appearance, with blurry or even “mottled” edges on objects (see image C below)
Figure 1: Axial images with varying levels of Iterative Reconstruction. Note the increase noise levels in image (a) due to no IR, and the extreme reduction of noise in image (c) with 100% IR.
Most radiologists do not like the appearance of a heavily-weighted IR image, so most protocols are setup to have a modest blend of IR and FBP that allows for about a 50% reduction in dose. This level is only a suggestion though. If you have scanned an image that you think is quite noisy, then it is possible to reconstruct with a higher IR that may be of higher diagnostic value. On the same hand, if you want to reduce the dose on one of your scans (for example a paediatric patient) and the radiologist is happy to allow more IR in the image, you can reduce the dose of the scan and increase the IR to compensate for the increase in noise in the raw data.
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 7: IV Contrast Flow and Timing
The timing of a scan after contrast has been injected can be critical in order to visualise the anatomy or pathology of interest. There are a few different ways we can time our contrast runs, each with their own unique pros and cons:
1. Post Injection Delay
This involves acquiring the scan at a preset post injection delay time after the contrast injection begins. Below are some typical CT Protocols that use Post Injection Delay:
-
Portal Venous Abdomen (65-70 sec PI)
-
Post Contrast Chest (35 sec PI)
-
Soft Tissue Neck (45-60 sec PI)
-
Nephrographic Phase (80-120 sec PI)
-
Post Contrast Brain (5 min PI)
Pros
-
Easier to setup and perform
-
Gives you more time to be in the room to monitor the injection
Cons
-
Does not account for natural cardiac output variation between patients
2. Bolus Tracking
This involves taking incremental low-dose axial scans of a region while the contrast is being injected. The operator can see contrast enter the body and can initiate the diagnostic scan either immediately after the contrast reaches a certain region, or a set delay afterwards. Most modern scanners have automatic triggering, where the operator places a ROI over a region (usually the aorta) and sets a HU threshold.
Pros
-
Accounts for natural cardiac output variation between patients
Cons
-
Extra radiation dose (minimal)
-
More chance of extravasation due to limited available time to be in the scanning room during contrast injection
3. Timing Bolus
This method uses a combination of the first two methods. We inject a small amount of contrast (10-15 ml) followed by a small saline flush and we monitor the contrast moving through a certain anatomical region using incremental low-dose axial scans. The difference between this and a bolus tracking is we don't trigger the scan when we see contrast arrive. Instead, we pause the monitoring and calculate the time it took for the contrast to arrive or peak at this region post injection (time to peak).
We then use this number to calculate two things:
-
The post injection delay time for the diagnostic scan
-
Contrast volume = Flow Rate x Scan Delay
Pros
-
Most accurate contrast enhancement and phase timing
-
More accurate contrast volume
-
Less beam hardening artefact from contrast remaining in the SVC
-
Less chance of significant extravasation incident
Cons
-
Extra contrast volume used (10-15ml for timing bolus, however this is often mitigated in the main scan)
-
Extra radiation dose (minimal)
-
Extra planning required
What Delay to Set
Have you ever wondered why certain protocols have specific contrast delay times? Well it all has to do with the time different organs take-up contrast once it enters the body via IV injection. The flowchart below may help you understand what delay time is best for different body parts:
.jpg)
Figure 1: IV Contrast flowchart for different body parts. Note the blue lines indicate venous flow and the red lines indicate arterial flow.
CT Champion Course
The CT CPD Series is made up of excerpts from the CT Champion Course - a self-paced, online course covering EVERYTHING you need to know about CT Radiography.
If you're enjoying the content, head over to the CT Champion Course page to find out what it can bring to your career.
Topic 8: CT Radiation Dose Measurement
Due to the unique acquisition method of CT, with a tightly collimated beam making multiple passes along the z-axis, specific units of measurement have been developed to make radiation dose monitoring for CT more relevant and accurate.
CT Dose Index (CTDI)
This is essentially an overall, averaged-out absorbed dose for the entire scan, determined by the scanning factors set in the protocol (mAs, kVp, pitch, etc.). Because it measures absorbed dose, the units are in Grays (or mGy for the levels we work with in CT). CTDI provides an estimate for deterministic effects based on calibrated scans of different-sized phantoms. These measurements have been indexed in multiple ways over the years, with each change an improvement for having a more accurate representation of the human habitus:
-
CTDI100: linear measure of dose distribution over a 100 mm long pencil ionisation chamber (outdated)
-
CTDIw: weighted average across a single slice, with ⅔ of the dose counted around the periphery and ⅓ in the centre (outdated)
-
CTDIvol: takes CTDIw one step further by taking into consideration variations in dose distribution along the z-axis. CTDIvol = CTDIw / Pitch (currently used)
CTDIvol is dependent on pitch. For a pitch of tighter or wider than 1:1, the CTDI is multiplied by the ratio of slice thickness to interval/overlap. So a pitch of 0.5:1 would have a higher CTDIvol than a pitch of 1:1, and a pitch of 1.5:1 would have a lower CTDIvol (this is assuming no other scanning factors are changed).
Dose Length Product (DLP)
DLP (mGy.cm) = CTDIvol x scan length
DLP takes into account both the scanning factors in your protocol as well as the scan length. Therefore it is the most accurate measurement to determine the dose for each individual examination.
Calculating Effective Dose from DLP
Absorbed dose (Gray) does not account for the type of radiation or the radiosensitivity of the body region, so therefore does not provide an accurate representation of risk. To do this, we need to convert the DLP (absorbed dose, mGy.cm) to effective dose (mSv), which accounts for both of the abovementioned factors. We achieve this by multiplying the DLP by a weighting factor specific to the area of anatomy scanned. The table below provides you with a variety of weighting factors (Source: NRPB-W67 - 2005):
Let’s look at some examples:
-
Adult CT Brain - DLP 700 mGy.cm
-
Eff Dose = 700 x 0.0021 = 1.47 mSv
-
-
Adult CT Abdomen - DLP 450 mGy.cm
-
Eff Dose = 450 x 0.015 = 6.75 mSv
-
You can see that although a CT Brain has a higher DLP than CT Abdomen, the risk is higher for the CT Abdomen because of the higher radiosensitive organs being exposed.
Diagnostic Reference Levels
DRLs provide a reference point for national CT doses across Australia. If your doses for a particular protocol are routinely above the DRL, then it is suggested that you review this protocol and aim to reduce the dose.
You can view the current DRLs here.
.png){kind=small}