Sunday 29 January 2012

Protocol for the QA of Computed Radiography Systems
Commissioning and Annual QA Tests
This document describes a series of tests to assess CR plate and reader performance. The tests are intended to detect artefacts and monitor image quality and sensitivity. The tests are split into the following categories,
             - commissioning tests
             - annual QA tests.       
All the tests described should be performed on all available reader systems.
KCARE have data from performing tests on all manufacturers CR systems and are available for advice and information.
1 Commissioning Tests
List of equipment
·        Tape measure
·        Adhesive tape
·        1.0 mm Copper filtration (>10 x 10 cm)
·        1.5 mm Copper filtration (>10 x 10 cm)- for Agfa only
·        0.5 mm Copper and 1mm Aluminium filtration (>10 x 10 cm) –for Kodak only
·        TO20 threshold contrast test object
·        Resolution test object (e.g. Huttner 18)
·        M1 geometry test object or lead ruler
·        Contact mesh
·        Ionisation chamber
·        Small lead or Copper block (~5 x 5 cm)
·        Steel ruler
In all tests described the unique plate identification code should be recorded.
These tests should be performed using an x-ray unit that has recently passed QC tests. In particular, the accuracy of the kVp selected for detector dose indicator consistency and calibration should be tested.
It may be possible to undertake some of these tests on a review workstation depending on the processing tools available (which will be dependent on the manufacturer). However certain tests require the use of the higher quality reporting workstation.

1.1 Dark Noise
Purpose:  To assess the level of noise inherent in the system
 a)   Erase an image plate and without making an exposure read it using the following parameters
   Text Box: Agfa:  S=800, 
examination type - ‘System Diagnosis’
processing – ‘Flat Field’

Fuji:  Readout mode – ‘Fixed’
  S = 10000
  L = 1

Kodak: Mode – ‘pattern’

Konica: Readout mode – Fix











b)      Examine the images visually for uniformity and record the detector dose indicator value for Agfa (SAL – at centre of plate) and Kodak (Exposure Index).
c)      Record a mean pixel value using region of interest analysis (for systems not offering ROI analysis see appendix for details of how to measure a mean pixel value).
d)      If possible either archive or print the image for future reference.
Tolerance: For Agfa and Fuji systems a uniform artefact free image should be expected.  Kodak systems add a collector profile to the image to compensate for non uniform collection efficiency across the place. This results in series of bands appearing across the image. Agfa systems should have an SAL < 100. For Fuji the pixel value should be <280. For Kodak the EI value should be <80 for GP plates and <380 for GP plates. For Konica a pixel value >3975 should be expected.
1.2 Dosimetry
Purpose: To measure receptor doses required for later tests 1.3,1.5, 1.6, 1.7 and 1.11
a)      Position an ion chamber at ~1.2 m from the focus (see figure 1) and at least 30 cm above the table (record the actual distances). Collimate to the ion chamber.




b)      Expose the chamber at 70kVp with 1.0mm of copper in filtration at the tube head. The mAs should, by ‘trial and error’, be set such that the inverse square law corrected dose to the table level is approximately 10mGy
c)      Record the measured dose and repeat twice.
d)      Under the same beam conditions determine the mAs required to deliver a receptor entrance air kerma of 1mGy, 4mGy 12mGy and 50mGy
e)      If a Fuji system is being tested determine the mAs to deliver a receptor entrance air kerma of 10mGy at 80kVp with no filtration in the beam.
If a Kodak system is being tested determine the mAs to deliver a receptor entrance air kerma of 10mGy at 80kVp with 1mm Al and 0.5mm Cu filtration at the tube head.
1.3 Linearity and System transfer properties
Purpose:­ To establish the relationship between receptor dose and pixel value so that this relationship can be corrected for in tests 1.4 and 1.7. Also to establish that the indicated exposure (calculated from the detector dose indicator) responds linearly to increases in dose.
a)   Place a 24 x 30 cm cassette on the table at ~1.50m (as described for test 1.2). Set the field to just cover the cassette. Mark the corners of the cassette on the table with transpore, so that the cassette can be easily repositioned.

b)   Expose a plate at 70kVp with 1.0mm copper at the tube head to deliver a dose of order 1mGy as measured in test 1.2.
c)   After a minimal fixed time delay (e.g.1 to 5 mins), read the plate as described below.
Text Box: Agfa:  S=200, system diagnosis/flat field processing

Kodak:  Pattern mode body part.

Fuji:  semi-auto, L=1 or 2

Konica: QC S-value, E and F processing turned off








d)         Record the detector dose indicator value.
Text Box: Agfa:  SAL (draw an region of interest covering at least 10000 pixels at                        the centre of the image

Kodak:  Exposure index (EI)

Fuji:  Linearity mode (S=200)

Konica: Fix mode








e)   Record a pixel value from the centre of the image.
§         For Agfa systems the SAL values obtained from ROI analysis on the review workstation should be used.
§         For Fuji, Konica and Kodak systems the images should be transferred to reporting workstations to use ROI analysis tools if available.
f)    Repeat for doses of order 4mGy, 12mGy and 50mGy.
g)   Plot a graph of pixel value versus receptor dose using a graph plotting package (e.g. Microsoft excel). Obtain the equation of the trend-line for this graph (i.e. the pixel value as a function of receptor dose). This equation is the system transfer properties (STP) equation and is used for making corrections in tests 1.4 and 1.7. An equation of the form
dose =  f (pixel value)
 where f is some arbitrary function is required.
Tolerance: For all images the ratio, k, of indicated exposure to exposure should not differ by greater than ±10% from the mean k value. The trend-line plotted in excel should have an R2 fit value >0.95. There is no tolerance for the STP equation. However the pixel value to dose relationship should be a simple relationship (e.g. log, linear or square root). For systems evaluated by KCARE the following has been found.
Manufacturer
STP Relationship
Agfa
Square root *
Fuji
Logarithmic
Kodak
Logarithmic
Konica
Logarithmic
* For the Agfa system there is a square root relationship between SAL values and dose. The relationship between raw data pixel values and dose however was logarithmic for systems evaluated by KCARE
1.4 Erasure cycle efficiency
Purpose:  To test that minimal residual signal (ghosting) remains on a plate after readout and erasure.
a)      Position a plate on the table at ~1.5 m. Set a 10 cm x 10 cm field and position a piece of attenuating material (e.g. Copper or lead) at the centre of the CR plate. Expose at 80kVp, 25mAs with no filtration.
b)      Read the plate (the readout parameters are not important).
c)      Re-expose the plate with a 9 cm x 9 cm field centred on the same point on the plate with no attenuating material in place, using 80kVp, 0.5mAs and no filtration.
d)      Read the plate using the following parameters.


Text Box: Agfa:  S=200, 
examination type - ‘System Diagnosis’
processing – ‘Flat Field’

Kodak: Mode – ‘Pattern’

Fuji:  Readout mode – ‘Semi Auto’
  L = 1 or 2

 Konica: Semi fix
    g=1


e)       










Set a very narrow window and adjust the level. Visually inspect the image for any remnant of the previous image (look for both the attenuating material and the position of the collimators). If a remnant is visible, use region of interest analysis to quantify the difference in pixel value between the ghosted and unghosted areas. 
¨      For Agfa systems the SAL values obtained from ROI analysis on the review workstation should be used.
¨      For Fuji, Konica and Kodak systems the images should be transferred to reporting workstations to use ROI analysis tools if available.
The ROI values should be used to calculate indicated receptor doses using the STP equation established in test 1.3.
Tolerance:  If no evidence of ghosting is found from visual inspection of the images then the test is passed and there is no need to perform ROI analysis. There should be <1% (remedial) difference between the STP corrected pixel values in the ghosted region and the surrounding areas. A suspension level of  <5% is set.
1.5 Detector dose indicator calibration
Purpose:  To assess the accuracy of the plate exposure values calculated using exposure indicators.
a)   Position a 24´30 plate on the table as described for test 1.3


Text Box: CR system Filtration Tube Voltage (kVp)
Agfa 1.5mm Cu 75
Kodak 0.5mm Cu +1mm Al 80
Konica info not known info not known
Fuji none 80
b)   Expose the plate to a known dose of ~10 mGy using the kVp and filtration listed below (use mAs found in test 1.2).  
c)   Read the plate out as described below
Text Box: Agfa: no delay between exposure and readout, S=200, system diagnosis/flat field processing and linear sensitomitry. 

Kodak: A 15 minute delay between exposure and readout, readout on Pattern mode body part.

Fuji: A 10 minute delay between exposure and readout, readout using semi-auto, L=1 or 2

Konica: Details of the Konica calibration protocol are not available











d)  Record the detector dose indicator, and calculate the indicated exposure using the equations given below.
Text Box: For Agfa systems the indicated exposure, EAgfa , in mGy, for a 200 speed readout is given by 

EAgfa = 5.90 ´ 10 –6 ´ SAL2     (1)

For Kodak systems the indicated exposure, EKodak , in mGy, is given by

¬ , where     (2)

For Fuji systems the indicated exposure, Efuji , in mGy, is given by 
       (3)
For Konica systems the equation linking to dose and S value is not available





















e)         Repeat twice and take a mean value of the indicated exposures.
Tolerance: The indicated exposure should agree with the measured exposure within 20%.
1.6 Detector dose indicator consistency
Purpose:  To assess the variation of sensitivity between plates, and set a baseline for monitoring system sensitivity for future QA testing
a)      Place a 24 x 30 cm CR cassette on the couch and set up as described for test 1.2/1.3 (see figure 1) and with 1.0mm Cu filtration.
b)      Expose the plate at 70kVp to give a known dose of ~10 mGy. The dose to the plate calculated from inverse square law corrected ion chamber measurements should be recorded (see test 1.2)
c)      Read the plate as described for test 1.5.
d)    Record the detector dose indicator, and calculate the indicated exposure using equations 1-3. Repeat twice for the same plate.
e)      Calculate the indicated exposure using equations given in test 1.5
f)        Repeat this test for all plates for acceptance testing(making only one exposure to each plate). It is helpful at this point to identify a plate that has a detector dose indicator in the middle of the range for future QA.
Tolerance: The variation in the calculated indicated exposures should not differ by greater than 20% between plates. The measurements repeated on the same plate should be used to lay down a baseline for future QA tests. Al images should be inspected for gross artefacts
1.7 Uniformity
Purpose:  To assess the uniformity of the recorded signal from a uniformly exposed plate. A non-uniform response could affect clinical image quality.
a)      Expose a plate as described for test 1.6 but with half the mAs.
b)      Rotate the plate through 180o about the vertical axis and re-expose using the same parameters (this should largely cancel out the non uniformities due to the anode heal affect).
c)      Read the plate as described for test 1.3.
d)      Visually inspect all images obtained in test 1.3, 1.5 and 1.6 for uniformity and artefacts. Likely artefacts include dust on the plate or readout optics, and scratches on plates. 
e)      The uniformity of the image obtained in 1.7b should be assessed using region of interest analysis (ROI ) if available, to measure the mean pixel values in positions a-e, as indicated in figure 2 below (i.e.at the centre of the image, and at the centre of each of the four quadrants of the image). The size of ROI should be of order 10000 pixels.
For Agfa systems the SAL values obtained from ROI analysis on the review workstation should be used.
For Fuji, Konica and Kodak systems the images should be transferred
to reporting workstations to use ROI analysis tools if available





















f)        The five values obtained from ROI analysis should be used to calculate five indicated receptor dose values using the STP equation obtained in test 1.3
Tolerance: The images should not have obvious artefacts. The ratio of the standard deviation of the 5 STP corrected ROI values to their mean (the coefficient of variation) should be less than 10%.
1.8 Scaling errors
Purpose:  To assess the accuracy of software distance indicators and check for distortion.
a)      Position the M1 test object directly on the centre of a CR cassette at > 150 FDD.
b)      Expose at 50-60 kVp with no filtration and 10mAs.
N.B. A lead ruler could be used in place of the M1 test object. If so 2 exposures should be made with the ruler placed in first the scan direction then the subscan direction. 
c)      Read out plate using processing as for test 1.3.
d)      Using the distance measuring software tools measure the dimensions (x and y) of five central squares in both fast and slow scan directions. Calculate the aspect ratio x/y. For Agfa systems the review workstation software can be used. For Fuji, Kodak and Konica systems the images should be transferred to the reporting workstation to use distance measuring software tools if available. If images are reported from film then they should be printed at full size. Distances can then be measured with a ruler. 


Reposition the test object over the edge of the plate as indicated in figure 3 and repeat steps b and c
f)        Along the edge of the plate measure the horizontal (x1) and vertical (y1) sizes of two squares as indicated in figure 4. Calculate the aspect ratio x1/y1.
g)      If possible download the image as a DICOM file. Open the image using a DICOM viewer such as Santeviewer. Hold the curser over a corner of a square in the grid. Record the position within the image (i.e. the x and y coordinates). Move the curser to the corner of the square of the grid 10cm from the first corner in the x direction. Record the coordinates again. Calculate the pixel pitch, p(mm)=100/n, where n =number of pixels covering 10cm of the grid. Repeat for the y direction. This test is only necessary on commissioning. Compare the pixel pitch to that stated by the manufacturer. The difference should be no greater than the estimated measurement error.
















Tolerance: The measured distances x and y should agree within 3% of the actual distances at the centre of the plate and 5% at the edge. All calculated aspect ratios should be within 1.00 ± 0.03 at the centre of the plate, or 1.00 ± 0.05 at the edge.
1.9 Blurring
Purpose:  To test for any localised distortion or blurring of the image.
a)      With the contact mesh in placed on the cassette at >150cm FDD, expose at 50-60 kVp, fine focus, with no filtration and 10mAs.Read the plate as described for test 1.3.
b)      Visually inspect the image for distortions. If distortion occurs clean the plate and repeat.
      c)      Repeat for at least two other plates.
d)      Repeat with a fine mesh if available.      
Tolerance: No blurring should be present. If blurring is present on all plates this suggests the reader is at fault, whilst imperfections in individual plates may also lead to blurring. If blurring remains on a region of a plate after cleaning it should not be used clinically.
1.10 Limiting Spatial Resolution
Purpose:   To test the high contrast limit of the systems ability to resolve details.
a)   Place a general purpose cassette on the couch with the Huttner test object positioned at its centre aligned at 45o to its edges.
             NB A Huttner test object with line spacings up to 8 lp.mm-1 may be required.
b)      Set 50-60 kVp, fine focus, and expose the cassette using 10mAs.
c)      Readout the plate using the following parameters
Text Box: Agfa:  S=100, 
examination type - ‘System Diagnosis’
processing – ‘Flat Field’

Fuji:  Readout mode – ‘fix’ with L=2 and S=200

Kodak: Mode – ‘Pattern’ with raw data and no edge enhancement

Konica: QC S-value, E and F processing off










d)      Adjust the window level and magnification to optimise the resolution. Score the number of resolvable groups of lines from the screen. Look up the corresponding resolution. The image should be scored at a magnification of order x 5. If this facility is not available on the review workstation then images should be transferred to the reporting workstation for scoring. 
e)      Repeat the measurement twice with the resolution test object placed at a slight angle to the first the lateral or then the longitudinal axis.
f)        Repeat this process for all available image pixel pitches (nb different plate sizes will often default to being scanned at different pixel pitches).
g)      If possible either archive or print the image for future reference.
Tolerance: These measurements should be used to set a baseline for future QA tests. Print or save the images for future reference, if possible. Comparable images are available for most systems through KCARE.
N.B. The limiting resolution should be expected to approach the Nyquist limit. At 45o the Nyquist frequency is defined by Ö2/2p where p is the pixel pitch. The measured limiting resolution may be limited by display when scoring from a review workstation, particularly if no zoom or limited zoom facilities are available. 
1.11 Threshold Contrast Detail Detectability
Purpose:  To monitor image quality by assessing the visibility of low contrast details.
a)   With the tube, plate, and 1.0mm copper filtration in the same positions as for the sensitivity tests, place the TO20 (or equivalent) test object on the plate. Collimate down to the size of the test object.
b)   Set 70 kVp and an mAs to deliver ~4 mGy. Read the plate using the following parameters.
Text Box: Agfa:  S=200, 
examination type - ‘System Diagnosis’
processing – ‘Flat Field’

Fuji:  Readout mode – ‘semi-auto’ with test/sensitivity GA=1

Kodak: Mode – ‘Pattern’ with raw data and no edge enhancement

Konica: QC S-value, E and F processing off










c)  Ascertain whether clinical images are most commonly viewed soft or hard copy. If they view hardcopy, adjust the window to optimise the visibility of the details, ensuring that background noise is perceptible, and print the image out on the largest film size. View the image on a masked light box, and score each detail size using fixed distance viewing (<1m). If images are viewed softcopy, score them on a reporting workstation optimising window and level settings for each detail size.
d)      Calculate an image quality factor, IQF,
                                                                                (4)

where  
HT(A) = threshold contrast detail index values calculated from the image,
HTref(A) = threshold contrast detail index values calculated from a reference  image of a system known to be in good adjustment
            D = the dose to the image plate
            Dref = the dose to the image plate for the reference image
            n = the number of details in the test object.
d)      Repeat this test for two other imaging plates and also for a single plate at exposures of ~1mGy and ~12mGy.
e)      If possible either archive or print the images for future reference.
Tolerance: The results of this test are used to set a baseline for future QA tests. Results could be compared to those from other similar systems if available.
1.12 Laser beam function
Purpose:  To assess laser beam scanline integrity and jitter
a)      Place a steel ruler slightly angled to the subscan direction on a large cassette.
b)       Expose at   ~70 kVp, 150cm FSD and an mAs to deliver an incident exposure of ~50mGy. Read the plate as described for test 1.3
c)      Using the software magnify the image x10. This will usually require the image to be viewed from a reporting monitor. Select a narrow window width such that the image appears largely polarised to black or white. This should allow the edge to be easily differentiated from the background. Laser beam jitter can be evaluated by examining the edge of the ruler on the image.
Tolerance: The edge should be continuous across the full length of the image. Stair step characteristics should be uniform across the length of the image. Regions of over or undershoot of the scan lines indicate a timer or laser beam modulation problem.
1.13 Moiré Patterns
Purpose:  To test for the presence of Moiré pattern artefacts caused by grids.
a)      Place a CR cassette in the bucky such that the scan lines are vertical to the gridlines. The cassette should be 1.5m from the focus, and the collimation should cover the whole plate.
b)      Expose at 70 kVp using the AEC with 1.0 mm of copper in the beam, and the grid in place. 
c)      Read the plate as described for test 1.3.
d)      Visually inspect the image for Moiré line pattern artefacts.
e)      Repeat with the CR cassette positioned in the bucky such that the scan lines are horizontal to the gridlines.
f)        Repeat for all buckies and grids that may be used with the CR system, including any grids used in mobile radiography.
Tolerance: No Moiré patterns should be visible. If Moiré patterns are visible with a particular grid, it should not be used with the CR plates. The cause of Moire patterns may be the failure of the motion of moving grids or insufficient grid density.

2 Annual QA tests
The following routine QA tests should be performed approximately annually
            1.2       Dosimetry (only for 4mGy and 10mGy with 70kVp and 1.0mmCu)
1.4       Erasure cycle efficiency
1.6       Detector dose indicator consistency/sensitivity (for 1 plate of each size)
            1.7       Uniformity
            1.8       Scaling errors
            1.9       Blurring
            1.10     Limiting resolution (45o only)
            1.11     TCDD (only 4mGy).
The tests should be performed as described in the previous section. Table 1 below summarises the relevant remedial levels where these are different to those described for commissioning
Test
Remedial Level
detector dose indicator consistency (sensitivity)
baseline ± 20% exposure equivalent
limiting resolution
baseline ± 20%
TCDD (Quality Index)
baseline ± 30%
It should be noted that TCDD and limiting resolution are subjective measures. Some effort should therefore be made to train scorers to score to similar thresholds.
Appendix A – Measuring a mean pixel value using a Fuji CR system
Reduce the window width to 1, so that the image has only pixels appear as either one of just two levels, black or white. Adjust the level until approximately half the pixels are black and half are white. This level value is the mean pixel value of the image.
References
[1] Draft Report of Task Group #10, American Association of Physicists in Medicine, Acceptance Testing and Quality Control of Photostimulable Storage Phosphor Imaging Systems, August 1998
[2] British Institute of Radiology, ‘Assurance of the quality in the diagnostic imaging department’, 2001, ISBN 0-905749-48-0
[3] IPEM draft CR QC protocol 
[4] Samei E, Seibert JA, Willis CE, Flynn MJ, Mah E, Junck KL, ‘Performance evaluation of computed radiography systems’, 2001, Med.Phys. Vol28 (3) p361-371.

No comments:

Post a Comment