Topics covered in this article include:

• Supported Phantoms
• Scanning for CBCT
• Varian CBCT
• Elekta XVI CBCT
• Measurements and Plots
• Recommended Tests
• Detailed Discussion of Measurements
  • Low Contrast
  • Modulation Transfer Function (MTF)
  • Sensitometry
  • Slice Thinckness
  • Spatial Linearity
  • Uniformity

Supported Phantoms

The Catphan® Phantom supports phantoms in the Catphan line of phantoms in the 500, 600, and 700 series of phantoms supplied for radiation therapy and diagnostic imaging QA. For radiation therapy, in particular, the following phantoms are supported:

• Varian Linacs: Catphan® 504 , Catphan® 604 
• Elekta Linacs: Catphan® 503

Phantom Rotation

• Center of Phantom - The center of the phantom described is the pixel coordinate in the center of the phantom (in pixels).
• Rotation - Degrees of phantom rotation around the z-axis.
• Tilt - Degrees of phantom rotation around the x-axis.
• Yaw - Degrees of phantom rotation around the y-axis.

Scanning for CBCT

The automatic analysis service performs best if the phantoms are scanned according to the standard protocols laid out by the LINAC manufacturers.

• Place the phantom case on the gantry end of the table with the box hinges away from the gantry. It is best to place the box directly on the table and not on the table pads. Open the box, rotating the lid back 180°.
• Remove the phantom from the box and hang the Catphan® from the gantry end of the box. Make sure the box is stable with the weight of the phantom and is adequately counterweighed to prevent tipping.
• Use the lasers (or linac crosshairs) to make sure the Catphan® is centered on the first lateral height dot and center dot that are proximal to the gantry.
• Use the level and adjusting thumb screws to level the Catphan®. The re-check the Catphan® centering on the first lateral height dot and center dot.
• Move the Catphan® in towards the gantry so that it is aligned with the 3rd alignment dots from the front of the phantom.
• Perform CBCT scan.

Varian CBCT

Image Owl supports all CBCT tests suggested by Varian provided in their Installation Product Accectance guidelines, outlined below. Catphan® 504 and Catphan® 604 versions are automatically identified and analyzed.

Scanning Modes

The following scan modes are used to demonstrate basic CBCT operation, functionality, and conformance.

• Standard Dose Head (full fan)
• Pelvis (half fan)

Phantom Setup

Position the phantom on the front edge of the couch top, using the storage box as the holder.

Align the phantom as follows:

• VRT Axis: Wall lasers intersecting the side reflective dots on phantom
• LAT Axis: Overhead or Sagittal laser intersecting the top reflective dots on phantom
• LNG Axis: Middle of the Low Contrast module CTP515 aligned to wall vertical laser

Before continuing, rotate the gantry 360º around the Catphan® phantom to make sure there are no collision possibilities.

Scanning Test Method

1. With the Clinac in CLINICAL mode, launch Treatment on the 4DITC and verify the PRO values (from the LVI) are displayed in the Treatment application. 
2. Open the test patient CATPHAN plan from the Treatment Queue, and mode up the CBCT setup field. Use DICOM RT mode and perform a Machine Override if the patient is not imported into the OIS and/or not available in the Treatment Queue.
3. Close the OBI application and launch the CBCT application.
4. Open a New Patient and enter a New Patient ID, using the ID names shown in the setup table for the applicable scan mode.
5. Select Acquire new scan and follow the on-screen instructions to acquire the desired scan mode.

Verification Tests

The full set of DICOM files from each scan mode can now be uploaded to Catphan® QA for analysis.

Elekta XVI CBCT

Catphan® QA includes all Catphan® tests from the ELEKTA CAT guidelines.
Automatic module identification is performed when a full phantom upload (or a set including the CTP404 module) is provided.
For partial phantom uploads (which do NOT include the CTP404), DICOM fields must be set as suggested in the ELEKTA CAT guidelines to provide automatic module identification.

• The system has only been tested on data with 120kV, 40mA, 40ms as suggested in the ELEKTA CAT guidelines.
• For robust results, please use a reconstructed FOV that just covers the phantom, 200-300mm

Recommended Tests

TG142 Summary Data

To include TG142 summary data for CBCT in analysis reports select measurements from the Cone Beam CT reports subsection of the Monthly Imaging QA [Tg142 Table VI] section when building your template. 

If you are using the Catphan® 503 phantom supplied with Elekta Linac's choose the TG142 summary data items from the ELEKTA CAT (CBCT and kV planar) reports subsection of Detailed Catphan Analysis.

Recommended Catphan 504 Tests

The recommended detail CBCT test selection for the Catphan 504 is shown below:

Recommended Catphan 604 Tests

The recommended detail CBCT test selection for the Catphan 604 is shown below:

Recommended Catphan 503 Tests

The recommended detail CBCT test selection for the Catphan 503 is shown below:

Detailed Discussion of Measurements

Low Contrast

CTP515 Low Contrast Module 

The low contrast targets have the following diameters and contrasts:

Supra-slice target diameters       Subslice target diameters

2.0 mm                                        3.0 mm

3.0 mm                                        5.0 mm

4.0 mm                                        7.0 mm

5.0 mm                                        9.0mm

6.0 mm

7.0 mm

8.0 mm

9.0 mm

15.0 mm

Nominal target contrast levels

0.3%

0.5%

1.0%

Low Contrast Measurements

Low contrast performance in CT imaging is a difficult subject. Clearly it is related to the ability of a "system" to distinguish between two objects or regions with similar CT number or X-ray attenuation. It is known to depend on statistical noise levels, contrast of the signal, and the size of the signal. In many cases, small signals (in the 3-7 mm diameter range) with contrast levels of some 3-5 HU are typically difficult to discern even in relatively uniform backgrounds. However, it is well reported that the contrast level (C) and the diameter of the target (D), when combined as a product approaches a constant (C*D≅constant). Thus, a plot can be made of contrasts (C) that might be found at a given diameter (D). 

Because of the difficulty of the subject, much thought went into the construction of the Image Owl low contrast algorithm. The original method, used during the beta release of the service was based on the Rose Model. This calculates a C-D diagram based on contrast, diameter, and noise levels of the scan. Specifically, from this model, a target was considered "detectable" when the CT# contrast is considerably greater than the noise (2-5 times). However, working with David Goodenough, Ph.D., a new method was developed. 

The previous method calculated low contrast values of measured contrast and pixel noise based on a contrast to noise ratio greater than 2. Simply, using known target locations, a target was considered detectable if the CT# contrast was 2 times the noise. The method the service now employs, estimates noise from the standard deviation of the mean HUs of two rows of circles of each target size in the background.

 In the current method for low contrast calculation, at each target size (2-9mm and 15mm), an inner and outer row of ROIs (circles) are generated in the background as shown above. The mean HU for each circle is calculated and the standard deviation (SD) calculated from the set of means. If the standard deviation from the outer and inner circles deviates more than 10%, the algorithm uses the set with the lower standard deviation. A detection level for each target size is calculated, with a perceptibility factor of 4, as 4*SD. 

Rose Model

The Rose Model represents probably the simplest model for detection of circular targets in a relatively uniform background (other than the statistical noise fluctuations seen in medical images). Strictly speaking, the model applies to "white" noise, meaning the noise excursions are uncorrelated from point to point (pixel to pixel). This is not necessarily true in CT, but nevertheless works reasonably well.

In the Rose Model, the eye (or the computer in this case) puts a circle (ROI of Region of Interest) over a target with the same diameter as the target and estimates the mean signal intensity of the target. This is a visual or numerical integration of the CT numbers in the target as averaged over the target. This value is then compared to mean intensity values in identical sized circles in the background region where only noise is present. A test of significance is made to determine how many times the mean signal intensity would exceed the expected fluctuation in the regions where only noise is present. The best case usually requires the mean signal intensity to be at least two times the expected noise deviation in white noise, and probably twice that for CT correlated noise.

Consider the case of Signal (S) and Background Level (B); with CT# of signal, CT_S; and CT# of background, CT_B.

For CT, contrast is CT_S - CT_B; and the CT noise is σ(H). Then signal contrast to noise ratio is CT_S - CT_B/σ(H).

To estimate noise excursions in signal sized targets, we note that for white noise, the relative noise standard deviation decreases by the square root of the number of pixels in the profile. If the single pixel noise standard deviation is "σ " then the profile (area) noise is approximately σ (H)/√N; where N is the number of single pixels needed to obtain the full area of the circular target. 

Consider a profile through a 5mm, 0.3% target in an ideal scenario. Here we estimate background as 50H; ∴0.3% target is 53H. Remember in CT, 1% change from water represents 10H. 

Now consider the same profile, but with single pixel noise at approximately 2H present in the profile.

If a typical pixel dimension is ≈0.5mm, then in a 5mm diameter target we may have π(5²)/4 mm² or ≈20mm² area. This is about 80 pixels, or 20/0.5². If then the single pixel noise is 2H is approximately 2H/√80, or 0.2H , for the area standard deviation noise. In this case, a 5mm target at 0.3% contrast should be seen quite well because the ration is approximately 15:1. This is well above the 2:5 ratio suggestion for detection with white noise and reasonably above the 4:10 ratio for correlated CT noise, which behaves as if the white noise level is approximately doubled. 

In a more normal (regular dose scan) technique with a typical 20cm phantom, where the single pixel noise is more like 4H, the ratio now becomes approximately 7.5:1, getting closer to the 4:10 limits of detection suggested for correlated CT noise. Moreover, a 3mm target with 0.3% contrast (a signal area about 1/3 that of the 5mm target with N ≈26) would have 0.7H effective noise for the area standard deviation and contrast to noise ratio of only about 4:1. This would not reasonably pass the 4:10 detection ratio, and therefore not be considered detectable. 

References

1. D.J. Goodenough, Psychophsical perception of computed tomography images. Radiology of the Skull and Brain; Technical Aspects of Computed Tomography, T.H. Newton and D. G. Potts, Eds, Vol. 5, pp. 3393-4021. Mosby, St Louis (1981).
2. D.J. Goodenough and K.E. Weaver, Factors related to low contrast resolution in CT scanners, Computerized Radiol. 8, Vol. 5, 297-308 (1984).

Modulation Transfer Function (MTF)

MTF Catphan® Modules

CTP528 with bead point source

CTP682 with bead and wire point source

 

Point Spread Function (PSF)

Pixel values around point source

The point spread function (PSF) describes the response of an imaging system to point source. In the case of the Catphan® QA, the point source is either a bead or wire. The image to the left shows the pixel values (HU) around the point source, with the highest values at the top of the color scale. The PSF can be thought of as the extended area that represents an unresolved object. The values from the PSF are used to derive the line spread function (LSF).

Line Spread Function (LSF)

The line spread function (LSF) is obtained by averaging the PSF values in the x and y directions. Once the averaged x and y values are symmetrized, the two directions (x and y) are averaged together to derive the LSF.

The above illustrations show how by summing the columns (y-axis) of numbers in the point spread function (PSF) the line spread function (LSF) for the x-axis is obtained. 

Modulation Transfer Function (MTF) 

MTF is the most commonly used method of describing spatial resolution capability. The MTF curve is used to graphically represent a system's ability to pass information on to the observer. The MTF curve results from the Fourier Transform of the LSF (described above) data. 

A CT system's ability to accurately resolve an object varies according to the size, the spatial frequency, of the object. As objects become smaller, they are more difficult to accurately resolve on a CT image. The MTF scale is from 0 to 1, with a value of 1 having the object reproduced exactly and a value of 0 having no image reproduced. The MTF curve graphs the MTF against the object size measuring a scanner's spatial resolution capabilities. Certain factors, including pixel size, field of view (FOV), slice thickness, and kernel, affect spatial resolution. Thin slices and smaller pixel size reduce volume averaging and improve resolution. Some of these factors, obtained from the DICOM header, are reported by the Catphan® QA to aid in MTF comparisons.

Nyquist Frequency

The Nyquist frequency reported on the MTF curve, represents the point at which the object cannot be accurately resolved. The Nyquist theorem, as applied to CT, tells us that because an object may not lay entirely within a pixel, the object should be two times the pixel size to increase the likelihood of being resolved.  

Information on the Nyquist frequency can be found here.

Sensitometry

CT or Hounsfield Numbers

Users of CT systems are often surprised when the CT number of a given tissue or substance is different from what they expect from previous experience. These differences do not usually indicate problems of a given CT scanner, but more likely arise from the fact that CT numbers are not universal. They vary depending on the particular energy, filtration, object size and calibration schemes used in a given scanner.

One of the problems is that we are all taught that the CT number is given by the equation: CT# = k(µ - µw)/µw, where k is the weighting constant (1000 is for Hounsfield Scale), µ is the linear attenuation coefficient of the substance of interest, and µw is the linear attenuation coefficient of water. Close review of the physics reveals that although the above equation is true to first order, it is not totally correct for a practical CT scanner. In practice, µ and µw are functions of energy, typical x-ray spectra are not monoenergetic but polychromatic, and a given spectrum emitted by the tube is “hardened” as it is transmitted (passes) through filter(s) and the object, finally reaching the detector. More accurately, µ=µ(E), a function of energy. Therefore: CT#(E) = k(µ(E) - µw(E))/µw(E)

Because the spectrum is polychromatic we can at best assign some “effective energy” Ê to the beam (typically some 50% to 60% of the peak kV or kVp). Additionally, the CT detector will have some energy dependence, and the scatter contribution (dependent on beam width and scanned object size, shape, and composition) may further complicate matters. Although the CT scanner has a built in calibration scheme that tries to correct for beam hardening and other factors, this is based on models and calibration phantoms that are usually round and uniformly filled with water, and will not generally match the body “habitus” (size, shape, etc.). The situation is really so complicated that it is remarkable that tissue CT numbers are in some first order ways “portable”!

In light of the above we can examine a parameter of CT performance, the “linearity scale”, as required by the FDA for CT manufacturer’s performance specifications. The linearity scale is the best fit relationship between the CT numbers and the corresponding µ values at the effective energy Ê of the x-ray beam. The effective energy Ê is determined by minimizing the residuals in a best-fit straight line relationship between CT numbers and the corresponding µ values.

 In review, we will encounter considerable inter and intra scanner CT number variability. CT numbers can easily vary by 10 or more based on kVp, slice thickness, and object size, shape, and composition. There is some possibility of the use of iterative techniques and/or dual energy approaches that might lessen these effects, but certainly CT numbers are not strictly portable and vary according to the factors listed above. 

Please note: The CT number measurements for the individual materials are the median of the measurements from the input slices.

Mass Attenuation Coefficient Table

sensitometrydata.xlsx

On the worksheet found at the link above are mass attenuation coefficients for sensitometry materials used in Catphan® phantoms. Data is provided for selected energies from 20 keV to 20 MeV. Contributions from different interactions are given as well as totals both with and without coherent scattering effects. The values were obtained from the NIST XCOM database using our best knowledge of material compositions. The data is subject to change pending new information.

Sensitometry Targets

CTP401 (Catphan® 500) contains: LDPE (low density polythylene), Acrylic, Teflon®, Air

CTP404 (Catphan® 503, 504, 600) contains: Polystyrene, LDPE, PMP (polymethylpentene), Air, Teflon®, Delrin®, Acrylic, and a vial for Water

CTP682 (Catphan® 700) contains: Teflon®, Bone 50%, Delin®, Bone20%, Acrylic, Polystyrene, LDPE, PMP, Lung Foam #7112, Air, and a vial for Water

The targets range from approximately +1000H to -1000H.

The monitoring of sensitometry target values over time can provide valuable information, indicating changes in scanner performance.

Nominal Material Formulation and Specific Gravity

Electron Density and Relative Electron Density

 ¹Z_eff, the effective atomic number, is calculated using a power law approximation.

² For standard material sensitometry inserts, The Phantom Laboratory purchases a multiple year supply of each material in a single batch. Samples of the purchased material are then measured to determine the actual specific gravity. The specific gravity of air is taken to be .0013. For custom cast materials the specific gravity of each cast batch is noted and supplied with the phantom. The Lung #7112 is a foam, and while it is purchased in large batches, its density varies through the batch. For this reason the lung numbers may have a greater variation.

³ These are minimum and maximum measured values from a sample of 94 scans using different scanners and protocols. The Bone 20% limits are not taken from actual measurements but are scaled from measurements taken using an insert with a slightly different composition from an actual Catphan®700 Bone 20% insert. HU can vary dramatically between scanners and imaging protocols. Numbers outside this range are not unusual. Water was not measured so nominal values of +/- 7 HU are given.

⁴ Relative Electron Density is the electron density of the material in e/cm³ divided by the electron density of water (H₂O) in e/cm³.

Slice Thickness

Slice Thickness from Bead Ramps

To illustrate how the bead ramps are used, the following illustration shows both a 1mm and 2mm slice going through a bead ramp. You may note that as the slice thickness increases, the peak CT value for the beads decrease. This is because as the slice thickness increases, the bead’s effect on the CT number of the voxel decreases, due to volume averaging. Presuming the slice thicknesses are accurate, the peak signal over background in a 1mm slice should be double that of the peak signal over the background in the 2mm slice.

 

Please note in this example 0.25mm spaced slice ramps are used, rather than the 1mm spaced slice ramps. The methodology for both would be basically the same, however for thinner slices the use of the fine ramps improve measurement precision. Below we use a 1mm slice scan to illustrate the use of a profile made from a line running through the bead images on the scan. 

When we use a profile line through the beads, there will be peaks at each of the bead locations and these will be separated by 0.25mm from each other.   Thus for example, for the 1.0mm slice width we measure about four bead spacings at the Full Width at Half Maximum (FWHM). Multiplying the four bead spacings times the y axis increment 0.25mm per bead yields a 1mm slice width.       

Another method for counting beads would be to measure the maximum CT number of the beads. This can be done by adjusting the window width to 1 and raising the level until the beads disappear and noting the peak level. Next, do an ROI of the area adjacent to the ramp to get a number for the background. Keeping the window width at 1, raise the level to half between background and peak (half maximum) and count the beads.
We can make this somewhat more analytic by noting the following.  If we hand-draw, or use a mathematical “best fit” bell shaped curve (Gaussian) to the data points, you will notice that the peak CT number for the 1.0mm slice is about 650 H and the baseline is about 50, leaving a net value of about 600H between the peak value and the baseline.  Thus, ½ the (net) maximum value is 300H + the baseline of about 50H so we draw a line across the 350H ordinate (Y) value and measure the length of the line that spans the two FWHM points at, in this case, 350H.  When measuring the FWHM of the curve it is important to realize that due to scaling and translation variables the scale of the FWHM length needs to be defined. This is done using the distance between the individual bead peaks in the profile whose absolute separation is known (.25mm for fine ramps and 1mm for course ramps). For example for the fine ramps divide the FWHM by the distance between bead peaks and multiply by .25mm.    

Slice Thickness from Wire Ramps

The 23° wire ramp angle is chosen to improve measurement precision through the trigonometric enlargement of 2.38 in the x-y image plane.
To evaluate the slice width (Zmm), measure the Full Width at Half Maximum (FWHM) length of any of the four wire ramps and multiply the length by 0.42(tan23º): (Zmm) = FWHM * 0.42 
To find the FWHM of the wire from the scan image, you need to determine the CT number values for the peak of the wire and for the background. To calculate the CT number value for the maximum of the wire, close down the CT “window” opening to 1 or the minimum setting. Move the CT scanner “level” to the point where the ramp image just totally disappears. The CT number of the level at this position is your peak or maximum value. To calculate the value for the background, use the region of interest function to identify the “mean” CT number value of the area adjacent to the ramp. Using the above CTvalues, determine the half maximum: First calculate the net peak... (CT # peak - background = net peak CT #) Calculate the 50% net peak... (net peak CT # ÷ 2 = 50% net peak CT #) Calculate the half maximum CT number... 
(50% net peak CT # + background CT # = half maximum CT #)

Additional Methods for Measuring Slice Thickness

Note: These methods are not used by the Image Owl algorithms. 
A ssp of the bead(s) can be used to measure slice thickness (see CTP528 section for additional information).
Sagital and coronal slices through the beads can also be used to measure the axial slice width. In this case measure the z axis length at the full width at half maximum of a bead image to establish the slice thickness. However this tecnique is limited in precision z axis of the voxels.
The volume averaging effect on the net peak CT number of the bead can be used to approximate additional slice thickness measurements after measuring one slice’s thickness by using the following equation: 
w = slice width of additional slice thickness. 
npvm = net peak value of the bead in the measured slice width 
msw = slice width of the measured slice
npva = net peak value of the bead in the additional slice widthw = (npvm / npva)*msw 
Note: Net peak value = (CT# of the bead) - (CT# of the background)

Spatial Linearity

Spatial Linearity of Pixel Size Verification

The spatial linerarity is the reported estimated pixel spacing. The distance between the spatial targets is measured (the small teflon and air plugs) and from the known distance the real pixel spacing is calculated.

These 3mm diameter holes are positioned 50mm on center apart. By measuring from center to center the spatial linearity of the CT scanner can be verified. Another use is to count the number of pixels between the hole centers, and by knowing the distance (50mm) and number of pixels, the pixel size can be verified.

CTP404 Module Target Spacing

Please note: The Teflon pin is used for identification and orientation only. The ability to change the Teflon pin position enables organizations with more than one Catphan® phantom to identify their phantoms by images of the first section.

Uniformity

The image uniformity module is cast from a uniform material. The material’s CT number is designed to be within 2% (20H) of water’s density at standard scanning protocols. The typically recorded CT numbers range from 5H to 18H. This module is used for measurements of spatial uniformity, mean CT number and noise value.

The precision of a CT system is evaluated by the measurement of the mean value and the corresponding standard deviations in CT numbers within a region of interest (ROI). These measurements are taken from different locations within the scan field.

The Image Owl algorithm provides vertical and horizontal line profiles across the uniformity module. The points on the plot are an average over 5 columns of pixels. The plot is then smoothed further to reduce noise before the fitted curve is completed. To calculate the uniformity index for the horizontal and vertical profiles, the maximum and minimum y-axis CT values (HU) from the fitted curve are entered into the following equation: 1 - (CTmax-CTmin) / (CTmax+CTmin)
The closer a value is to "1", the more uniform the image. 

 The phenomenon of “cupping” or “capping” of the CT number may indicate the need for calibration.

In the graph below, the upper limit is determined by adding 40 HU to the Mean CT number of the Center ROI. The lower limit is determined by subtracting 40 HU from the Mean CT number of the Center ROI.