QC in the ACR CTAP

Douglas Pfeiffer, MS, DABR

Boulder Community Hospital
Introduction

QC is essential for the consistent production of quality images

Documentation is essential

Active, not passive

Does not need to be time-consuming

Manufacturer Programs
Manufacturers required to provide QC phantom and
recommendations [21 CFR Ch. 1 §1020.33(d)]

Contrast scale

Noise

Nominal tomographic section thickness

Spatial resolution - high and low contrast

Mean CT number - water or reference material

The Rest Might Be A Lie!
Renovated CT Submission
CTAP submission WILL be changed

Streamlined process

No alignment test

No reconstructed scan width test

Scan phantom using clinical protocols

Adult head, abdomen

Pediatric head, abdomen

Renovated CT Submission

Streamlined process…

Submit DICOM data on CD

MRI model

For now, dosimetry will remain the same

Low contrast transitioning to CNR

Spatial resolution going away

Siemens
Measures

Slice position

Slice thickness

CT number homogeneity

Noise

Water CT number

MTF

Table positioning

After alignment, scanner automatically scans phantom, computes, logs data

General Electric
Measures

Slice position

Slice thickness

Spatial resolution

Low contrast

Noise

Water CT number

Manual scan, calculation, documentation

Philips
Measures

Slice Thickness

Noise & Uniformity

“Impulse Response” (MTF)

Contrast Scale

Low Contrast Detectability

Spatial Measurement Accuracy

Manual scan of each module, automated analysis

Toshiba
Multiple phantoms

Measure

Slice Thickness

Noise & Uniformity

“Impulse Response” (MTF)

Contrast Scale

Low Contrast Detectability

Spatial Measurement Accuracy

Manual scan, manual analysis

Technologist QC
Image Quality

High contrast (spatial) resolution

Low contrast resolution

Image uniformity

Noise

Artifact evaluation We’re not

allowed to
Alignment light accuracy
specify any less
Slice thickness than what’s in
the standards!
CT number accuracy and homogeneity

Display devices

Frequency determined by QMP

Personnel
Best to identify a single individual

Greater consistency

Improved sensitivity to problems

QC must be completed, regardless of individual

Appropriate training must be provided

QC Notebook

QC policies and procedures

Data forms for each test

Area for comments and communications

Service engineer should also utilize the notebook

Data reviewed at least annually by physicist

Supervising physician should also review

Procedures

Daily

Recommendations provided

Alternatives acceptable

Manufacturer specific

Automated systems

As recommended by Qualified Medical Physicist

ACR Accreditation Phantom

RMI Model 464

Uniformity
Alignment Artifacts
Slice thickness
CT number accuracy
Action Limits
Must be established for each test

Over time, limits may be tightened based on scanner performance

Medical physicist should establish and make recommendations at

review
Alignment Light Accuracy
Position phantom

If manufacturer’s (MFG), use manufacturer’s holder

If ACR, place on table or use optional stand

Use alignment lights

Scan phantom
MFG: Use manufacturer protocol
ACR: Adult head technique
Thin slice: ~1 mm
Alignment Light Accuracy

Action limits
MFG
Use manufacturer limits
ACR
± 2 mm

Correction within 30 days

CT Number Accuracy and Noise

Warm up scanner per manufacturer recommendations

Perform air calibrations per manufacturer recommendations

Position phantom

If manufacturer’s (MFG), use manufacturer’s holder

If ACR, place on table or use optional stand

Use alignment lights

CT Number Accuracy and Noise
Scan phantom
MFG: Use manufacturer protocol
ACR: Adult head technique
Recommend both helical and axial
Consider alternating modes each day
Place ROI on image
400 mm2
Same location each time
Same slice if multi-slice
CT Number Accuracy and Noise
Action limits
MFG
Use manufacturer limits
ACR
Water value: 0 ± 5 HU
Noise: Limits must be established by the medical physicist
based on the scan protocol used
Schedule service if either value exceeds action limits 3 days in a
row or 3 times in one week
Artifact Analysis
Position phantom
Scan parameters (assume approx. 16 cm water)
Axial
120 kVp
350 mAs
Maximum number of slices possible
Want to check every data channel
320 slices?
Helical
120 kVp
350 mAs
Reconstruct to about 2.5 mm
Artifact Analysis

Visual analysis

Window width 100

Window level 0

Rings, streaks, lines

Record all findings

Artifact Analysis
Corrective action
While still sub-clinical
Re-run air calibrations
Often cover just a sub-set of scan conditions
May need to run several times
If still present
Determine if scanner can be used
Radiologist, medical physicist
Limited use?
Schedule service
Artifacts
Laser Printer QC
NEMA
Density Constancy
Making measurements on the TG18-QC or SMPTE test pattern, four
density steps. For dry development systems that are self-calibrated, this
test should be performed weekly. For wet processing systems, this test
should be performed daily.
Fixer Retention
A fixer retention test is required only for wet processing systems. Similar to
that for conventional film processing systems.
Low-Contrast Visibility
An image containing low contrast objects in a background of known
surrounding densities should be analyzed semi-annually to ensure that
low contrast objects are accurately reproduced.
Laser Printer QC
NEMA (continued)
Spatial Resolution
An image containing high contrast spatial resolution targets should be
analyzed semi-annually to ensure that fine details can be adequately
visualized.
Darkroom Fog
If darkroom used. The applicable MQSA regulation [21 CFR
900.12(e)(4)(i)] for conventional film processors must be followed.
Artifact Analysis
Images including large areas of uniform mid-density, such as the
AAPM TG18-UN or TG18-UNL, should be reviewed annually for such
artifacts.
Laser Printer QC
Very few wet lasers left in the field

Dry lasers generally have self-calibration feature

Is QC necessary?

YES!
Laser Printer QC
Case study
Kodak 8700 laser printer
Self-calibration performed on schedule
QC not performed
Got complaints regarding image quality
Sensitometric curve significantly different from same model
Internal densitometer had drifted and needed replacement
Recommend monthly QC
Laser Printer QC
Visual analysis
5%, 95% patches visible
Quantitative analysis
Measure 0%, 10%, 40%, 90% squares
Plot values
Action limits of ± 0.15 from operating level
 40%

10% 90%

0%
From the ACR
MRAP QC
Manual
Follow Up

Quality control must be active, not passive

Data must be analyzed immediately

All results must be logged or plotted

Corrective action must be taken promptly

Oversight

The QMP should review the QC data at least annually

Be visible

Be accessible

Data should be presented to the service engineer

Radiologists must be involved

Reporting artifacts
 Phantoms
Manufacturer
Designed specifically for unit
Automated testing
May not allow all “physics” measurements
ACR CT Accreditation
Good general, all purpose phantom
Direct comparison to ACR standards
Phantom Lab
“Gold standard”
Many advanced measurement capabilities
Physicist Tests
Alignment light accuracy
Alignment of table to gantry
Gantry tilt
Slice localization from scanned projection radiograph (localization
image)
Table incrementation accuracy
Slice thickness
Image quality
High-contrast (spatial) resolution
Low-contrast resolution
Image uniformity
Physicist Tests
CT number accuracy and linearity
Display devices
Video display
Hard-copy display
Dosimetry
Computed tomography dosimetry index (CTDI)
Patient radiation dose for representative examinations
Safety evaluation
Visual inspection
Audible/visual signals
Posting requirements
Contrast Scale

CS =
(µ r − µw )
(CTr − CTw )
where r, w refer to reference material and water, respectively

Must accurately know µr

µr and µw are constant

If CTr and CTw constant, CS constant

Not used
Gantry-Table Alignment

Fixed at installation

Earthquake? MVA?

Determine exact midline of cradle

Verify that sagittal alignment light remains within ±5 mm of cradle

midline as couch translates
Scatter Survey

Should not change with time

Worst case

Large dosimetry phantom

Maximum radiation beam width

High mAs, 140 kVp

Alignment Light Accuracy
Most important when biopsies are performed
Required Test Equipment
Phantom incorporating externally visible radiopaque fiducial markers or an image
center indication (ACR CTAP Phantom Module 1)
Test Procedure Steps
1. Using the alignment lights, carefully position the phantom to the radiopaque
markers in all three orthogonal planes.
2. Zero the table location indication.
3. Scan the phantom in axial mode, using a reconstructed scan width less than 2 mm
at the zero position. Use technique to allow accurate visualization of the fiducial
markers; for most phantoms, the Adult Abdomen technique works well.
4. Useful to scan also at ±0.5, ±1.0
Alignment Light
Accuracy
Data evaluation

±1 mm if
biopsies
±2 mm
otherwise
Scout Prescription Accuracy
Required Test Equipment
Phantom incorporating radiopaque fiducial markers or an image
center indication (ACR CTAP Phantom Module 1)
Test Procedure Steps
1. Scan the entire phantom in scout mode.
2. Magnify the image, if possible, and position a single cut at the
location of the radiopaque fiducial markers.
3. Perform an axial scan using a reconstructed scan width less
than 2 mm, or as thin as the scanner can produce in axial
mode.
Scout Prescription
Accuracy
Data Evaluation

±2 mm
Slice Thickness

Detected / Reconstructed scan width

Rarely an issue

Radiation beam width

Often an issue
Reconstructed Scan
Width
Required Test Equipment
Phantom with internal targets allowing determination of reconstructed image
thickness (Module 1 of the ACR CTAP Accreditation Phantom)
Test Procedure Steps
1. Align the phantom to the reconstructed image thickness determination targets
in the phantom.
2. Using zero table increment and techniques adequate to allow unambiguous
visualization of the targets (for most phantoms, 120 kVp, 200 mAs is
adequate), scan the phantom in axial mode using each reconstructed image
thickness used clinically
As many different data channel (detector) configurations should be used as
possible.
Reconstructed Scan
Width
Data Interpretation and Analysis

View the axial images collected above

Determine the reconstructed image thickness of each nominal

thickness tested

For the ACR CTAP Phantom, each line represents ½ mm

thickness

Count each line that is at least 50% of the brightness of the

brightest line
Reconstructed Scan
Width
Reconstructed Scan
Width
Count if half the brightness of the brightest

±1.5 mm
Radiation Beam Width
Required Test Equipment
External radiation detector (CR plate, self-developing film, electronic test tool)
Flat radiation attenuator (1/8” lead)
Test Procedure Steps
1. Place the radiation attenuator on the table, unless contra-indicated by the test
device being used.
2. Place the external radiation detector on the flat attenuator.
3. Adjust the table height so that the external radiation detector is at the
isocenter.
Scan using each unique N×T combination available, adjusting table position as
appropriate for the beam width being used.
Radiation Beam Width
 Radiation Beam Width

GE LightSpeed
Plus (4 slice)
±2 mm or 30% of N×T

1.25 5 10 15 20

4.7 8.2 12.2 16.6 21.1

Radiation Beam Width
Data Interpretation and Analysis
Using a method appropriate for the external radiation detector used,
determine the actual radiation beam width for each unique N×T combination.
For film and CR-based measurements, determination should be made at the
full width at half maximum (FWHM).
Precautions and Caveats
Many manufacturers have standards that are in excess of the criteria stated
below. It has been the experience of the ACR CTAP Physics Subcommittee
that most scanners can be calibrated to meet these tighter standards.
The above notwithstanding, scanners may exhibit over-beaming that can
impact these tolerances.
Table Travel Accuracy
Required Test Equipment
Phantom with two sets of external fiducial markers of known separation (ACR CTAP
Phantom Modules 1 and 4)
If possible, additional weight on the table to simulate the weight of a typical patient.
Test Procedure Steps
1. Using the alignment light, carefully position the phantom to the first set fiducial markers
in the axial plane.
2. Zero the table position indication.
3. Move the table to the second set of external fiducial markers.
4. Record the table position.
5. Translate the table to full extension and return to the first set of fiducial markers.
6. Record the table position.
Table Travel Accuracy
Data Interpretation and Analysis

Using the number recorded in Step 4, Compare the distance

between the fiducial markers as determined by the table travel
to the known distance.

Compare the number recorded in Step 6 to the zero position.

Precautions and Caveats

Some scanners have specific limitations on the extent of table

travel under which the performance specifications are valid.
Scanner-specific limitations must be noted.
Gantry Tilt
Required Test Equipment
“Ready Pack” film or CR image recording device
Protractor
OR phantom with radiopaque markers appropriately placed
Test Procedure Steps
1. Position the image recording device parallel to the sagittal laser, perpendicular to the scan
plane
1. It is helpful to use an acrylic slab
2. Expose the film at various gantry angles
1. Use ∼1 mm image width
2. Include +, - and 0
3. Process the image
4. Measure the angles with a protractor
Gantry Tilt
Alternate Test Procedure Steps
1. Position the phantom with the
markers on the cradle, centering it
to the markers at isocenter
2. Scan the phantom at the gantry tilt
angles prescribed by the markers
3. Observe the images, verifying the
visibility of the angle markers
1. If needed, rescan with the
gantry angle offset until the
markers are visible
Gantry Tilt

±3°
Low Contrast Detection
Required Test Equipment
Phantom incorporating low contrast targets of known contrast (Module 2 of the ACR CTAP
phantom)
Test Procedure Steps
1. Align the phantom.
2. Perform clinical scans covering the low contrast section.
1. Any Auto mA feature must be disabled
2. Use an mAs value appropriate for an average sized patient.
3. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 year old)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
Low Contrast Detection
Data Interpretation and Analysis
Visual Analysis
View each series and determine the slice providing the visually best low contrast
performance.
Adjust the window width and window level to optimize visibility of the low contrast
targets. On the ACR CTAP phantom, this is WW=100, WL=100.
Record the size and contrast of the smallest visualized target.
Numeric Analysis
Select the slice most central to the module containing the low contrast targets.
Place a Region of Interest (ROI) over the largest representative target and record
the mean HU value.
Place an ROI adjacent to the target and record both the mean HU and the HU
standard deviation.
target mean HU− background mean HU
Calculate the Contrast-Noise Ratio as CNR =
background HU std dev
Low Contrast Detection
Noise limited

Specific realization of the noise may hide a target

For ACR

OK to select any slice in Module 2 for low contrast

performance evaluation

If all 4 lower rods visible, count larger

Use best slice

Low Contrast Detection

6 mm
Low Contrast Detection
CNR measurements

Extremely sensitive to reconstruction algorithm

The specific algorithm used must be recorded for each clinical

protocol.
The Use of a Simple
Auto mA features must be disabled Contrast to Noise
Ratio (CNR) Metric to
Predict Low Contrast
Resolution
Failure to do this leads to a low mA value Performance in CT
Med. Phys. Volume 36,
Issue 6, pp. 2451-2451
Poorer performance than should be (June 2009)
Low Contrast Detection

CNR = 0.25 CNR = 1.06

High Contrast Resolution
Line pair pattern

MTF

Point response

Via line pair standard deviation per Droege

and Morin. A
practical method to measure the MTF of CT
scanners. Med. Phys. (1982) vol. 9 (5) pp. 758-
60
High Contrast Resolution
Required Test Equipment
Phantom incorporating high contrast targets of known resolution
(Module 4 of the ACR CTAP phantom)
Test Procedure Steps
1. Align the phantom
2. Perform clinical scans
1. Auto mA feature must be disabled
2. Use an mAs for average sized patient.
3. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 y.o.)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
High Resolution Chest
High Contrast Resolution

12
4 10

5 9

6 8
7
High Contrast Resolution
CT Number Accuracy

Why is this important?

Diagnosis

Bolus tracking

Treatment planning
CT Number Accuracy
Required Test Equipment
Phantom incorporating targets providing at least three different, known CT number values
including water (or water equivalent material) and air. (ACR CTAP Phantom Module 1)
Test Procedure Steps
1. Align the phantom
2. Perform clinical scans covering the CT number accuracy section
1. Disable any Auto mA features
2. Use an mAs value appropriate for an average sized patient
3. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 y.o.)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
4. Perform scans of the CT number accuracy section of the phantom with each kVp setting
available on the scanner. (Adult abd technique works usually)
CT Number Accuracy
Data Interpretation and Analysis
Select the image most central to the module containing the CT
number accuracy targets.
Adjust the window width and window level to optimize visibility of the
targets. On the ACR CTAP phantom, this is WW=400, WL=0.
In the image from the 120 kVp scan, place a circular ROI,
approximately 80% of the size of the target, in each target.
Record the measured CT number mean for each target.
For the images from each kVp scan, place a circular ROI,
approximately 80% of the size of the water target, in the water target.
Record the measured water CT number mean in each image.
CT Number Accuracy
All available kVp stations must be calibrated and tested.

The CT number accuracy targets in most phantoms are calibrated

only for 120 (or 130) kVp.

CT numbers other than water will vary with kVp

Need to know for bolus threshold setting

May be useful information for scanners used for RT

For ACR Phantom, use 200 mm2 ROI

CT Number Accuracy

Material CT # Range
Water -5 to +5 HU
Air -970 to -1005 HU
Teflon (bone) 850 to 970 HU
Polyethylene -107 to -87 HU
Acrylic 110 to 135 HU
CT Number Uniformity
Required Test Equipment
Phantom incorporating a uniform region, preferably water or water-equivalent
Test Procedure Steps
1. Align the phantom.
2. Perform clinical scans covering the CT number uniformity section.
1. Disable any Auto mA features
2. Use an mAs value appropriate for an average sized patient.
3. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 y.o.)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
4. Perform scans of the CT number uniformity section of the phantom with each kVp setting
available on the scanner. (Adult abd technique works usually)
CT Number Uniformity
Data Interpretation and Analysis

Select the image most central to the CT number uniformity

module.

In each image, place a circular ROI, approximately 400 mm2

(ACR phantom), at the center of the phantom.

Place similar ROIs at 12:00, 3:00, 6:00 and 9:00, one ROI
diameter in from the periphery.

Record the measured CT number mean and standard

deviation.
CT Number Uniformity

ROIcenter - ROIn # 5
Dosimetry

Not just for patient dose estimates

Generator calibration

Tube condition

Dose characterization
Dosimetry
Acceptance testing
Air
Output every kVp
Output for every filter
] Characterization and
potential dose calculation

Phantom

]
CTDI100 for every kVp Characterization,
CTDI100 for every filter calibration and potential
dose calculation
CTDI100 vs. mA
Compare scanner CTDIvol with measured CTDIvol, every filter with matched
phantom
±10%
CTDIvol for adult head and body, ped head and body clinical protocols
Compare to reference levels Less than ref. level
Dosimetry

Annual testing

Air

Output at every kVp ±5% of baseline

CTDIvol for four clinical protocols

Compare to reference levels Less than ref. level

Verify scanner CTDIvol

±10%
Artifact Evaluation

Arguably the most important evaluation to be done

Most likely to impact clinical scans

Most likely to have deficiency

Artifact Evaluation
Required Test Equipment
Phantom incorporating uniform section of adequate length for all data channels to be simultaneously used.
(ACR CTAP Phantom Module 3)
Test Procedure Steps
1. Align the phantom.
2. Perform clinical scans covering the uniformity section.
1. Disable any Auto mA features
2. Use an mAs value appropriate for an average sized patient.
3. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 y.o.)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
4. One scan in axial mode using approximately Adult Head techniques but with thin (approximately 1 mm)
images using the maximum number of data channels available on the scanner.
5. Repeat 4 for all kVp stations.
6. Repeat 4 for all bow tie filters
Artifact Evaluation
Required Test Equipment
Phantom incorporating uniform section of adequate length for all data channels to be simultaneously
used. (ACR CTAP Phantom Module 3)
Test Procedure Steps
1. Align the phantom
2. Perform clinical scans covering the uniformity section of the phantom. Use an mAs value appropriate
for an average sized patient. The scans performed should include at least
Adult Head (average)
Pediatric Head (1 y.o.)
Adult Abdomen (average)
Pediatric Abdomen (50 lb, 20 kg; 5 y.o.)
3. One scan in axial mode using approximately Adult Head techniques but with thin (approximately 1
mm) images using the maximum number of data channels available on the scanner
4. Repeat 3 for all kVp stations
5. Repeat 3 for all filters at 120 kVp
Artifact Evaluation
Data Interpretation and Analysis
Select the image most central to the uniformity module
Adjust the window width and window level to optimize visibility of the
phantom material. On the ACR CTAP phantom, this is WW=100,
WL=0.
Inspect each slice for evidence of artifacts, including specifically
streaks and rings, cupping, or capping
Some scanners will demonstrate a noticeable bright halo just inside the
phantom border, due to automatic corrections for a skull that it assumes
will be present. This artifact may be excluded from consideration.
Artifact Evaluation

QuickTime™ and a
Photo - JPEG decompressor
are needed to see this picture.

Clinically relevant - address immediately

Artifact Evaluation

Radiology 2005; 236:756 –761

Artifact Evaluation
Protocol Review
Review at least five clinical protocols, including

Pediatric Head (1 y.o.)

Pediatric Abdomen (50 kb, 20 kg; 5 y.o.)

Adult Head

Adult Abdomen

High Resolution Chest

Pay specific attention to

kVp

Mas
Protocol Review
Firm rules for clinical imaging are difficult to establish. The ACR has set several practice
standards:
Reconstructed scan width for standard Adult Head and standard Adult Abdomen should
not be less than ~5 mm.
Pitch for Pediatric Abdomen should not be less than 1.
Several other rules of thumb should be kept in mind:
On a multi-slice scanner, the largest number of images for the chosen reconstructed
scan width should be used, as this improves dose efficiency. For example, 5 mm x 4
slices is up to 30% more dose efficient than 5 mm x 2 slices in axial mode with no image
quality penalty.
The facility may wish to be able to reconstruct to thinner in addition to the standard
scan. N×T should allow for this.
Lower kVp settings should be used for pediatric scans and scans in which contrast
media is used.
Protocol Review
High Resolution Chest (HRC) protocol should incorporate a very sharp
reconstruction algorithm (GE recommends the BONE algorithm).
HRC should be axial, thin slices separated by 10-20 mm. If HRC
image are extracted from a helical chest scan, it must be verified that
the chest scan is used appropriately for diagnosis.
Doses should be as low as reasonably achievable. For the two stated
head protocols and the two stated abdomen protocols, the CTDIvol
must not exceed ACR Pass/Fail levels and should not exceed ACR
reference levels.
No changes should be made without the knowledge and agreement
of the radiologist.
Conclusion

A well-designed QC program is essential to quality imaging

Focus on likely failure modes for the specific scanner

Technologist QC does not need to take a lot of time