1/9/2013

Digital Mammography Quality Assurance

Quality Control
All-encompassing management program used to
ensure excellence in healthcare through the
systematic collection and evaluation of data.
Patient scheduling, management techniques,
departmental policies and procedures, technical
effectiveness and efficiency, in -service education
and image interpretation

Advanced Health Education Center

Quality Control
Quality Control

Quality control is the part of the quality

assurance program that deals with Quality Control are simple checks that ensure
the digital system is operating to the standards
techniques used in monitoring and
that is was designed to do.
maintenance of the technical elements of
The system is designed to detect any changes
the systems that affect the quality of the in settings that could compromise image
image. quality, and deterioration in the equipment
Level I Noninvasive and Simple performance over a period of time.
Level II Noninvasive and Complex
Level III Invasive and Complex

How is QC broken down for FFDM Why Quality Control

Indirect Conversion/Direct Conversion Reduce exposure to patients and

Also per each manufacturer specific personnel
FDA approved QC manual per Manufacturer. Consistent image quality
Updated version # of QC Tests
Detect and correct for potential
State specific also
problems, before they impact image
quality

1
 1/9/2013

Why quality control? Universal QC for Digital Mammography

Determination of what is Normal Not currently available

Detection of what is Abnormal
Understanding of how to return to ACR is working on a universal manual FFDM
Normal from abnormal.
In particular, in FFDM, how do you know you Currently manufacturer specific
are seeing is what it is suppose to be.
We will discuss this at the end of class

Quality Control Tests-Other Modalities

900.12(e)(6) Note:

For systems with image receptor modalities other

than screen-film, the quality assurance program shall For new unit: Must use most current
be substantially the same as the quality assurance
version
program recommended by the image receptor
manufacturer, except that the maximum allowable For renewal unit: Can use older
dose shall not exceed the maximum allowable dose version (version used when tested
for screen-film systems in paragraph (e)(5)(vi) of this
previously)
section

Technologists Quality Control

Technologists Quality Control Procedures GE Procedures Lorad Selenia

1. Monitor Cleaning 1. Laser Printer Quality Assurance

2. Viewing Conditions for the RWS 2. Viewboxes and Viewing Conditions
3. Flat field and Image Quality Checks 3. Softcopy Workstation QC
4. Phantom Image Quality and CNR 4. Artifact Evaluation
5. Viewbox and Viewing Conditions Test 5. Signal-to-Noise and Contrast-to-Noise
6. MTF Measurement Measurements
7. AOP Mode and SNR Check

2
 1/9/2013

Siemens Mammomat Novation QC

Lorad Selenia

6. Phantom Image Detector Calibration-Weekly and as needed

7. Detector Flat-Field Calibration Artifact Detection-Weekly and as needed
8. Visual Checklist Phantom image quality-daily-image phantom and
score
9. Repeat Analysis
SNR and CNR Measurements-Weekly
10. Compression Printer Check-when clinical images are to be printed-
Daily
Repeat Analysis-quarterly
Compression force-semi-annually

Technologists Quality Control

MQSA Inspections and Selenia QC Manuals
Procedures Fischer Senoscan
1. Detector Calibration and Flat Field Test Inspectors were requiring facilitys with Hologic Selenia
2. Phantom Image Acquisition Test units that they must have and work from the most
current Selenia QC manual released by Hologic.
3. Phantom Image Quality Test This is a misunderstanding and not a requirement. It is
4. System Resolution/Scan Speed Uniformity not necessary to have a copy of the latest QC manual
5. Image Display Monitor(s) Test for each system. There should be no penalties to
facilities using the QC manual that was shipped with
6. System Operations their Selenia system or with subsequent system
7. Repeat Rate upgrades.
8. Compression
9. Printer

Basis for this Basis for this

To date, all revisions to the Selenia QC To date, none of the software

manuals have not increased any QC test upgrades have affected QC
performance standard or action level performance criteria or action levels.
from those stated in earlier versions. If If this were to change in the future,
this were to change in the future, Hologic
Hologic must notify the FDA and all
would first need to obtain FDA approval
and then notify all affected customers affected customers.

3
 1/9/2013

The facility can chose how to

Basis for this
perform the QC procedures:
New system or upgrade to existing system-the Use the QC manual provided with the
newest revision of the QC manual applicable individual Selenia system or system
to the model (tungsten or Molybdenum) is upgrade.
included.
If a site has multiple Selenia systems, different
revisions of the QC manual may be in place For the sake of simplicity choose to
based on the age of the systems and the standardize QC across all systems if using
status of system upgrades. the same model type (tungsten or moly).

GE Senograph DS
Hologic
Quality Control Procedures
If the facility has both tungsten and moly systems, they Same as done on GE Senograph 2000D
must use the QC manual applicable to their model
EXCEPT
All QC tests are done internally- there is not
Hologic has confirmed this understanding with the FDA
and if a facility is told by an inspector they must have mathematical calculations done by the
and follow the latest revision they should ask the technologist, just documentation of results.
inspector to call the FDA hotline for clarification

Prior to QC Procedures Monitor Cleaning-GE

The detector temperature must be stabilized. Frequency: Daily/days when clinical image acquisition
or reviews are planned
The unit must be turned on for a period of
time to get the detector temperature Viewing Conditions Check for RWS-
appropriate. GE

Frequency: Daily/days when

image reviews are planned

Objective: Ensure optimal

viewing conditions

4
 1/9/2013

Monitor Cleaning Monitor Cleaning

Do not use cleaning agents which attack the surface,

such as petroleum (mineral) spirits.
Daily
The front panel is extremely sensitive to mechanical
damage. Avoid all scratches, knocks, etc. Ensures good image review conditions
Do not apply the cleaning liquid directly to the Lightly dampen cloth with water/solution
monitor housing or screen. Clean with cloth or cleaning tissue to
Do not allow the cleaning liquid to enter the monitor remove dust, finger prints and other
housing; be sure to dampen the cloth sparingly marks
Record completion in book

Viewing Conditions Check For

Flat Field-GE
The Reading Rooms
Frequency: Weekly
Each room has a data form that is filled out by Objective: Five tests performed during Flat Field
the Medical Physicist for the room test.
configuration. 1)brightness uniformity, 2)high frequency
The medical physicist looks at the normal modulation (HFM), 3)SNR uniformity, 4)bad ROI,
5)bad pixel verification.
value of the ambient light in the room and
measures it. First day of week/ first thing in the morning- image
If changes are made in the room such as small receptor retains ghost images taken during the week.
lamp, new doors, room curtains, blinds, etc Technical factors-automatic
Two exposures

Brightness Uniformity High Frequency Modulation(HFM)

Checks the magnitude of Looks at the magnitude of

low frequency structures or high frequency structures
large shapes in defects or small defects

5
 1/9/2013

Signal-to-noise ratio (SNR) Bad Pixel Verification

Uniformity, which is the Measures the number of

variation of the SNR over pixels having signals much
the whole image different from those of
their neighbors and the
density of the bad pixels

The Ideal response of the Detector is a

Bad Regions Of Interest (ROI) uniform image with the following

Counts the number of No large or small defects

regions of interest No pixel having a signal much
containing more than higher or much lower than its
neighbors.
two bad pixels A uniform signal-to-noise ratio

If Test Fails look at the following

Artifact Evaluation

The compression paddle and grid cover have

been removed
No object except the Flat Field test object is in
the field TO MAKE SURE THAT THE IMAGE IS FREE
The collimator is open to the largest field size OF UNDESIRABLE ARTIFACTS
The tube arm angle is at O degrees (LORAD)
The Flat field test object is clean and free from
scratches or other imperfections
The surface of the image receptor is clean
The Flat Field test object fully covers the FOV
of the image receptor

6
 1/9/2013

Artifact Evaluation
Select the First Flat Field view from the examination screen
window on the Acquisition Workstation. Activate the Full Zoom/pan function
Acquired an exposure on the preview screen
Make sure to change the Center and Width numbers to make
it darker
Do two exposures with Mo Filter and the second in Rhodium
Filter or one with Tungston filter and second image with Ag.
Depends on what system you have.
Pan across the whole image to look for artifacts. Move
zoom/pan like your mowing the lawn.
When image displays record the mAs value

If you want to establish a Mean Pixel

Value, click the ROI in preview tools An average center number is around
and click User draw. Then choose 64 500 and the average width setting is
X 64 from the dropdown list. Click around 600. But make sure it isnt
the center of the flat field image and too dark or too light because there
the ROI Statistics dialog box will will be artifacts that are missed. So
appear. Take the Mean Pixel Value you want a dark gray look to the
with the first three numbers for a image
start for the Center Setting.

If artifacts are present rotate the acrylic 180 and

repeat procedure
If the appearance of the artifacts changes location Case Study for
between the two images they are in the phantom
and there is not a system problem. Flat field check
If the artifacts persist in the same location there is a
problem in either the x-ray system or the digital
receptor

7
 1/9/2013

DICOM Printer Artifact Evaluation Performance Criteria

Print a flat field pattern to the printer. Do not take a Artifacts traced to the digital image
flat field image is not appropriate for this test.
receptor or the x-ray unit shall be
Inspect the laser film for artifacts
eliminated by a qualified service
If multiple FFDMs only do this test from one of the
machines and use the same machine every time for engineer within 30 days of the test
consistency.
After you have looked for artifacts on the film, then
record on the form.

Siemens Artifact Detection Performance Criteria

Objective-to determine if the detector is dusty, No clinically relevant artifacts should be seen
damaged or has other artifacts
Insert collimator mounted plexi phantom (40 mm on the image
thick) Must be corrected prior to further
Insert compression plate simulator examinations
Choose procedure QC raw
28 kVp 90 mAs
Look at image for clinical relevant artifacts by
magnifying to full resolution
If the image has artifacts do calibration and repeat

Inspection News GE FFDM

GE Phantom
Phantom QC Requirements
Clarification of GE FFDM Phantom QC testing GE-does not intend for a facility to perform routine
requirements weekly phantom QC on devices that are not used for
QC manuals have phantom being done on image interpretation
both the AWS monitor and the printer weekly If a facility does its final interpretations using
FDA contacted GE to clarify the printer softcopy only, it does not have to perform routine
requirement if a facility only does soft copy weekly phantom test evaluation using the printer.
interpretation The facility must still perform a phantom QC test
prior to printing final interpretation quality images.

8
 1/9/2013

Phantom Image Quality on AWS, RWS,

and Printer and CNR Test-GE Phantom Image Quality

Frequency: Weekly after establishing the Objective: CNR (Contrast to Noise Ratio) is a
baseline for the Contrast-to-Noise Ratio (CNR) measure of the detectors ability to distinguish
test. between objects in an image and the image noise
Ensure adequate/consistent quality of images
Only after successful completion of the Flat acquired by the detector and displayed on the AWS
Field test. and RWS monitors and the printer.
Phantom Image quality test of the printer-run Contrast Resolution is the ability of an imaging
only after successful completion of the daily system to distinguish similarities of objects in an
QC test for the printer image.

Analog Phantom QC/Digital Analog Phantom QC/Digital

Phantom QC Differences Phantom QC Differences

In film-screen mammography imaging In GEs digital imaging the relative level of a

systems the phantom image test is signal or contrast to the image noise is the more
for the consistency of image contrast relevant measure of image quality.
as represented by the density difference Therefore, the measure of consistency
(DD) between the image of an of CNR is used as a replacement
added test object (4mm thick acrylic disk) for the measure of consistency of DD.
and the background density of the phantom.

CNR

B. Change in CNR Measurement The following events require re-establishment of

the CNRol:
Operating level for the CNR ratio measurement must replacement of the x-ray tube
be established, CNRol. replacement of the Mo x-ray beam filter
5 consecutive days to determine a 5 day average and replacement of the compression paddle
a CNR operating level. replacement of the phantom
Subsequent weekly measurements are compared to replacement of the anti-scatter grid
this operating level. replacement of the detector
re-calibration of detector gain

9
 1/9/2013

Why the raw image? CNR

Tests depend on using numerical values B. CNR measurement change continued

that are proportional to the amount of x- Open the acquired raw image, zoom factor of 1, adjust
rays detected. the WW and WL between 125 and 175 to achieve the
Raw images provide image numbers that best contrast/object detectablity.
behave in the above manner.
Processed images are good for viewing but
are useless for the numerical tests.
The values used are ones that best
represent the incidental x-ray beam and
the response of the detector when raw
image is used.

CNR
Now do the math:
Calculate the CNR as
(mean_background-mean_mass)/sd-background

Calculate the change in the CNR:

If the new CNR is smaller than or equal to the CNR operating
level (CNRol), then calculate :
Change in CNR=1-(CNR/CNRol)

If the new CNR is larger than the CNR operating level (CNR ol),
then calculate
Change in CNR=(CNR/CNRol)-1

Why use the number 1 in the

Example: formulas
CNRol = 3.92
Mean_mass = 921.26 Change in CNR=1-(CNR/CNRol)
Mean_background= 969.43
sd_background=12.01 Change in CNR=(CNR/CNRol)-1
(mean_background-mean_mass)/sd-background
969.43-921.26=48.17 = 4.01
So you end up with a whole positive number
12.01
Change in CNR=(CNR/CNRol)-1
4.01 - 1 = .02 change in CNR
3.92

10
 1/9/2013

Contrast to Noise Mean

Define contrast to be the signal An average of a group of numbers

difference between two tissues A or data points.
and B
CAB=SA-SB
We are assuming that SA > SB so that
contrast is always positive.

Action Limit:
The change in CNR must not exceed 0.2.
System fails- the source of the problem must be How do you score the Phantom
identified and corrective action taken before any
further examinations are performed.

Passes-proceed to next step...score the phantom

11
 1/9/2013

GE Senographe DS and Essential

Phantom IQ Test on The Printer
Phantom Scoring
Only on AWS Action Limit:

Fibers-4
Masses-3
Calcifications-3
less than above-failed
Identify source of problem
Corrective action taken before any further
examinations are performed.

Siemens Phantom-Daily Performance Criteria

Objective-to ensure that adequate image quality is Must score 5-4-4

achieved
Phantom-no disk Correction required before examinations if
Record mAs failed
Program 2-4.5cm breast technique
Score on monitor
If there is a problem send it to the review
workstation and printer and examine.

12
 1/9/2013

Signal to Noise Signal to Noise

Bandwidth-in computer networks it is a

The power ratio between a signal
synonym for data transfer rate-the amount
(meaningful information) and the
of data than can be carried from one point
background noise.
to another in a given time period (bits of
Both signal and noise power must be data per second).
measured at the same or equivalent points
in the system and within the same
bandwidth (width or range of frequencies SNR=the ratio of the mean pixel value to
that an electronic signal uses on a given the standard deviation of the pixel values.
transmission medium).

Standard Deviation Siemens SNR and CNR

In statistics-(the number you get for Objective-to assure proper functioning of

the standard deviation is considered the solid-state detector by evaluating the
statistics) a measure of how much SNR and the CNR of the detector
the data in a certain collection are Use the baseline values for SNR and CNR
scattered around the mean. for weekly testing consistency
.

Siemens SNR and CNR Performance Criteria

Choose procedure QC raw
Compress phantom SNR 40
Program 2-4.5 cm average breast technique
Select AEC sensor 2 at the AWS
Acquire phantom image
Draw ROI in largest mass and record mean value Deviation SNR 15%
Draw ROI in background and record mean value and standard deviation
Calculate:

SNR=
mean (bg) DC (offset-always 50)
std (bg)
Deviation CNR 15%
Calculate:

CNR= (mean_background-mean_mass)/sd-background

13
 1/9/2013

Signal-to-Noise and Contrast-to-Noise

Measurements-LoRad
SNR and CNR Lorad

Frequency: weekly Phantom image will appear on monitor.

Objective: To assure consistency of the digital image ROI over disk slightly smaller than the disk-record
receptor Mean Value (Mean)
SNR (Signal to Noise Ratio) compares the level of the Drag the previously drawn ROI next to the disk
desired signal to the level of background noise toward the chest wall-record the Mean Value (Mean)
phantom and disk and standard deviation (STD). Do not use the Signal-
18x24 compression paddle-on phantom as close to to-Noise ratio given by the ROI statistics box.
4.5 cm as possible
Use clinically used exposure factors: (i.e. Auto Filter) Accept Image..

The SNR shall be computed using the mean and

standard deviation values obtained from the ROI
next to the acrylic disk.
Now SNR is automatic with the SNR 1. Compute the SNR of the detector according to:
button on monitor
SNR = mean background - DC offset
std deviation

mean background and std background=the mean and standard

deviation obtained from the ROI statistics dialog for the ROI
next to the acrylic disk.
DC offset is a DC offset added to the detector signal and is equal
to 50.

2. Compute the CNR of the detector according to:

SNR and CNR
CNR =
mean background mean disk
std background Performance Criteria
mean disk= the mean value obtained from the ROI statistics
dialog for the ROI on the acrylic disk. The measured SNR must be equal to or
greater then 40.
3. Compute the deviation from the original CNR
measurements according to: The computed CNR must be within 15%
CNR base CNR measured
of the value determined by the medical
Diff = CNR base X 100
physicist
CNR base=the CNR base value established by the medical physicist during
acceptance testing of the digital detector
This is recorded in the Signal-to-Noise Ratio (SNR) and Contrast-to-Noise Ratio
(CNR) Control Chart
CNR measured is the new CNR computed in step 2.

14
 1/9/2013

Siemens SNR and CNR Siemens SNR and CNR

Choose procedure QC raw
Objective-to assure proper Compress phantom
Program 2-4.5 cm average breast technique
functioning of the solid-state Select AEC sensor 2 at the AWS
Acquire phantom image
detector by evaluating the SNR and Draw ROI in largest mass and record mean value
the CNR of the detector Draw ROI in background and record mean value and
standard deviation
Calculate:
Use the baseline values for SNR and
CNR for weekly testing consistency SNR=
mean (bg) DC (offset-always 50)
std (bg)

Calculate:

CNR= (mean_background-mean mass)/sd-background

Performance Criteria Noise

SNR 40 Radiographic noise or mottle
The unwanted random
Deviation SNR 15% (uncorrelated), nonrandom
(correlated), or static level in a
Deviation CNR 15% radiograph that has been given a
uniform x-ray exposure

Noise Quantum Mottle Noise SNR & DQE

The random spatial variation of x-ray The ratio of the magnitude of the image
absorbed in the image receptor signal to the noise
Detective quantum efficiency
Fewer x-rays = noise or SNR &
SNR transfer efficiency (good information
visibility of subtle contrasts
transfer)
Microcalcifications that can be the first High DQE good information transfer &
sign of cancer may not be visible in a more dose efficient
noisy or underexposed image Low DQE poor information transfer &
less dose efficient

15
 1/9/2013

SNR Detector Design Noise SDNR

Design & calibration of the detector & SDNR

electronics for adequate dynamic Signal difference-to-noise ratio
range & number of bits of digitization A measure of the difference between a
are essential to precisely record the signal & its background divided by the
entire range of x-ray intensities noise
transmitted by the breast Indicator of reliably depicting a structure
in the breast in the presence of noise
Radiation dose depends on desired SDNR

Viewbox and Viewing Conditions Spatial Resolution

Frequency:
Weekly
Objective:
To ensure good image review conditions by keeping the
High Contrast
viewboxes free of dust, finger prints, and other marks and the
viewing conditions optimized.
Procedure:
This test is not unique to digital mammography systems.
Follow accepted mammographic QC procedures
and action limits to complete this test.

Spatial Resolution Spatial Resolution

Loss most easily observed when imaging
The ability of an imaging system to allow 2 fine detail
adjacent structures to be visualized as
Speculations radiating from a mass
being separate, or the distinctness of an
edge of the image (ie. Sharpness) Microcalcifications

16
 1/9/2013

Spatial Resolution - Spatial Resolution

Qualitative Measurement Geometric Blurring

Achieved with a bar pattern of Minimized by using small FS for contact

alternating radio-opaque bars & imaging (e.g., 0.3 nominal size)
radiolucent spaces of equal width Minimized by using an even smaller FS for
magnification (e.g., 0.1 nominal size)
Determines limiting resolution in
Minimized by OID as much as possible
lp/distance or lp/mm
SID (e.g., 60-65 cm)

Spatial Resolution Detector Spatial Resolution Detector

Specific Blurring Specific Blurring

Occurs in x-ray converter material Direct flat panel detectors

Scintillator-based converters Voltage or electric field across the direct
First source of blur = spreading of emitted light conversion material must be adequate
within the scintillator material
Determined by
Material thickness
Crystal structure
Reflective & absorptive properties

Spatial Resolution Detector Spatial Resolution

Specific Blurring Motion Blurring

PSP resolution characteristics Caused by movement of the breast

Not determined by the emitted light spread during exposure
Spatial sampling is determined by
Minimized by
Size of the scanned laser beam on readout
Laser beam effective size (effective del) Short exposure time
determined by actual beam size & amount Compressing the breast
of scattering of the laser light within the
phosphor

17
 1/9/2013

Spatial Resolution Spatial Resolution

Motion Blurring Motion Blurring

kVp may be SSCCD - scan slot charge-coupled

Image processing compensates for device systems
contrast losses to the extent allowed Misregistration artifacts between
by the anatomy imaged before motion
Background noise occurs & that imaged after
Image SNR

Spatial Resolution Spatial Resolution

Geometric Blurring Motion Blurring
Contact imaging
Movement of the breast during
Minimized by using small FS (0.3 nominal
size)
exposure
Minimized by OID Minimized by a short exposure
SID (60-65 cm) time & compression
Use an even smaller FS for magnification
(0.1 nominal size)

MTF Measurement-GE MTF Measurement-GE

Frequency: Contrast is evaluated by measuring the

Monthly (only after successful completion of the Flat fluctuation of a bar pattern signal in a region
Field Test) of interest.
Objective: Monitor the contrast delivered by the The bar pattern provides a signal to the
detector detector, which is essentially alternating bars
Ensure contrast is adequate over the 0-5 lp/mm and spaces.
spatial frequency range by obtaining an estimate of The test determines how close to black the
the MTF (Modulation Transfer Function) values near signal at the position of the bars is and how
2 and 4 lp/mm. close to white the signal at the position of the
spaces is.

18
 1/9/2013

MTF Measurement-GE MTF Measurement-GE

Spatial resolution is the ability to see the
The closer the output signal is to difference in detail between adjoining
black and white, the more contrast objects, which is why we look at the spatial
frequency groups. Attention is given to both
there is in the image and the more low and high spatial frequency,
Contrast at low spatial frequencies, which we
variation is recorded in the region measure at 2 lp/mm, aids in detection of
masses and fibers.
of interest during the MTF Test. Contrast at high spatial frequencies, which we
measure at 4 lp/mm, aids in detection of
microcalcifications.
The greater the signal fluctuation, the greater
the MTF and the greater detectability.

MTFMTF Measurement-GE MTF Measurement-GE

100 micron => 5 lp/mm Procedure:

70 micron => 7.14 lp/mm Resolution bar pattern including spatial frequency groups of 2
0.1 and 4 0.1 lp/mm and a thickness of at least 0.1 mm of
54 micron / standard resolution => 9.26 lead.
lp/mm Positioning the pattern on the bucky and expose using the
27 micron => 18.5 lp/mm supplied technical factors.
Measurements made on raw image
Zoom image to 1, adjust the brightness and contrast for
Film screen 15 lp/mm and film screen mag is optimum visibility of the test object.
18 lp/mm
Using the ellipse tool, measurements are made.

MTF Measurement-GE MTF Measurement Test

Measurements:
2 lp/mm pattern
MTF is always expressed in %
4 lp/mm pattern
space material-mean_space
bar material-mean_bar
A line pair consists of two elements-a bar and a
space. The bar is the highly attenuating element and
the space is the low-attenuating element.

19
 1/9/2013

MTF Measurement Test MTF Measurement Test

Example:
Estimate the MTF expressed in % at 2 lp/mm:
2lp/mm
3335.74 (sd) X 222 = 740534.28 = 77%
mean space-mean bar
9864.99-295.81 = 9569.18

MTF Measurement Test What does the 222 represent in the formula?

A decision based on a dilemma. Mathematically it is

Example: [100 x x Square Root(2)/2].
Estimate the MTF expressed in % at 4 lp/mm:
4lp/mm It is based on a method of obtaining MTF estimates
1274.42 (sd) X 222 = 282921.24 = 29.5% from images of bar patterns.
mean space-mean bar Using a calculation with pi and a square root in it
9864.99-295.81 = 9569.18 would have perplexed the technologists and is a
waste of time for the tech to recalculate every time.
A simple constant (222) was decided to be used in
place of.

MTF Measurement Test CNR and MTF Measurement

Senograph DS and Essential
Frequency:
Weekly.
Action Limit:
The measurement must be made only after successful completion of the
The test is successful if: Flat Field Test
Objective:
The test is designed to check the consistency of the contrast to noise ratio
MTF 2lp/mm > 58% and (CNR) and to ensure that
MTF 4lp/mm > 25% contrast is adequate over the 0-5 lp/mm spatial frequency range by
obtaining an estimate of the MTF
(Modulation Transfer Function) values at 2 and 4 lp/mm.
If these results are not obtained, the source of the CNR measurement is done in two steps:
problem must be identified, and corrective action - Establishment of a baseline operating level CNRol
taken, before any further examinations are
performed.

20
 1/9/2013

CNR and MTF Measurement CNR and MTF Measurement

Senograph DS and Essential Senograph DS and Essential
1. Click on the QAP button on the right column of the Browser window. A 6. The following parameters are selected automatically:
list of tests is displayed. Rh/Rh/30kV/56mAs.
Select the CNR and MTF test. 7. Perform one exposure.
2. Enter or verify the reference of the IQST device (Serial Number or SN, 8. After the image has been captured, the results of the tests are
written on the side of the displayed:
device) on the AWS screen, then click Start . - The values of MTF at 2 lp/mm and MTF at 4 lp/mm.
Note: - The value of the change in CNR, computed as follows:
If the device reference entered is different from the previous one, you will Change in CNR = |CNR - CNRol| / CNRol
be asked if you want to where CNRol = the CNR Operating Level as described above.
restart the calibration process with this new reference. If CNRol has not been calculated yet, the change in CNR is computed as
3. Install the Bucky on the digital detector if it is not already installed. follows:
4. Remove the compression paddle. Change in CNR = |CNR - mean| / mean
5. Position the IQST device on top of the Bucky. where mean = the mean of the CNR values previously stored

AOP Mode and SNR Check-GE AOP Mode and SNR Check-GE

Frequency: Procedure:
Monthly A set of acrylic plates (minimum size 20 cm 20 cm)
allowing thicknesses of 25 0.1 mm, 40 0.1 mm
Objective: and 60 0.1 mm used with 5 deca newtons
Checks the following aspects of system operation: compression force
-correct choice of parameters in AOP (Automatic
Optimization of Parameters) mode
-correct level of SNR (Signal-to-Noise Ratio) in the
image

AOP Mode and SNR Check for GE AOP Mode and SNR Check for GE
Senograph 2000D Senograph 2000D
This test is done with a set of acrylic plates allowing
25+/-- 0.1mm and 40+/-- 0.1mm and 60+/-- 0.1mm Three exposures: one for each of the three
used with 5 deca Newtons compression force. thicknesses of acrylic in the field of view using the
AOP STD mode.
Record the exposure parameters after each
exposure.

Open each raw image for review and view it with the
default zoom (true size).

21
 1/9/2013

AOP Mode and SNR Check for GE

Senograph 2000D

Use the ellipse ROI tool:

Measure the mean value
Standard deviation, sd, of the image in the region
close to the chest wall edge and laterally centered.
Calculate the SNR:
mean= SNR
sd
Document results.

GE Senograph DS

and Action Limit:

GE Essential
AOP Mode Test successful:
Exposure parameters are in accord with the values
specified by the manufacturer.

If the system fails the test, the source of the problem

must be identified, and corrective action taken,
before any further examinations are performed.

GE

Example:
Calculate the SNR as mean/sd The Signal-To-Noise test is another
25mm mean = 865.86 sd = 11.42 measurement related to our ability to detect
= 75.81
objects in the image.
865.86
11.42 Signal refers to the average of the numerical
values in an area of the image. The signals in
40mm mean = 891.41 sd = 12.04 the image are the anatomical structures such
= 74.03 as glandular tissue, adipose tissue,
891.41 calcifications, and masses, which are
12.04 evaluated in interpreting a mammogram.

60mm mean = 809.81 sd = 12.25 = 66.10

22
 1/9/2013

Noise refers to the random variation of the

When the signal-to-noise ratio is
signal.
large, then noise does not obscure
This random variation can obscure the
clinical information the observer is trying to
the objects of interest in the image.
detect. When the SNR is small, objects and
The signal-to-noise ratio is a measure of the noise can be confused and the
relative strengths of the signal and noise. objects of interest; for example,
That is, the signal, or useful image masses and calcifications, may be
information, divided by the noise, or difficult to detect.
random information.

Manufacturers parameters in following table:

The SNR is also important because it sets

the limit on the amount of contrast
enhancement you can use to try to make
the objects easier to see. Such
enhancement generally increases both the
signal and the noise and reduces the SNR. The value of SNR must exceed 50
If the initial SNR is not sufficiently high, the
enhancement lowers the SNR to the point
where the contrast-enhanced noise
eventually obscures the objects of interest.

Visual Checklist-GE Compression Thickness Indicator

Frequency:
Monthly and after any service or maintenance on the
mammography system.
Objective:
To assure that the mammographic x-ray system
indicator lights, displays, and mechanical locks and
detents are working properly and that the system is
mechanically safe.

23
 1/9/2013

Remember.

The compression Thickness indicator

shall always be accurate to +/-- .5cm
from the actual thickness.

Image Display
No record forms needed for this
test, if any problems, consult the
medical physicist

Monitor Primary Display (RWS)

Display workstations used for official

Softcopy Display Monitors
interpretation of mammographic
images
FDA recommends primary display
monitors be cleared for FFDM by the
FDA
ACR strongly recommends only FDA -
cleared monitors

24
 1/9/2013

Monitor Primary Display (RWS) Monitor Primary Display (RWS)

Once a display has been purchased & Maximum luminance of grayscale monitors
calibrated, it should be tested regularly @ least 450 cd/m2
by the medical physicist to maintain Reflected ambient light from the display
compliance surface should be included in luminance
5 megapixel monitor (2,000 x 2,500 measurement
Pixel samples in the horizontal & Minimum of 8 bit luminance resolution (bit
vertical directions for portrait depth) is required
orientation) preferred Two-monitor portrait set up

Monitor Primary Display (RWS) Monitor Primary Display (RWS)

Ability to select image sequence & During readout all images should be viewed
display format (hanging protocols) at 1:1 or 100% size
Ability to accurately associate the patient Pixel size (or pitch) should be less than
& study demographic information with ~200microns
the images of the study performed Display device specifications should match
Eyeglasses specifically for viewing as closely as possible the acquisition matrix
distances (15 60 cm) size

Monitor Primary Display (RWS) Monitor Primary Display (RWS)

Window & level adjustment tools must

be available All images acquired in the study
Zoom (magnification) & pan (roaming) need to be fully accessible during
capabilities must be available interpretation
Rotation & Flipping tools are essential
Calculation & display of linear
measurements, ROI & pixel value
determination should be possible

25
 1/9/2013

Monitor Primary Display (RWS) Monitor Primary Display (RWS)

Clinically relevant technical parameters of the

acquired image data should be accessible Sufficient for viewing all types of CR/DR
mAs images
kVp
Bit depth
Exposure time
Matrix size
Exposure values to assess technique for
dose, quality, & feedback for technologists

Monitor Primary Display (RWS) Monitor Primary Display (RWS)

Reflections from ambient light sources

should be kept at a minimum
Optimize viewing conditions
Indirect & backlight incandescent lights with Control reading room lighting
dimmer switches rather than fluorescent Eliminate reflection on the monitor
Color tint should be uniform across the
display area
Lower ambient lighting level as much
Monitor pairs should be color matched from as feasible
the same manufacturing batch

Monitor Primary Display (RWS) Softcopy Display Monitors

Other Guidelines

Displays must be able to

Ambient lights should NOT be turned
Display mammography CAD marks when CAD is
off completely nor turned up implemented
completely Apply marks on the displayed image
20 lux is generally sufficient corresponding to all findings encoded in the
DICOM mammography CAD strured reporting
(SR) objects
Display images in true size

26
 1/9/2013

Softcopy Display Monitors

Display Device Calibration Check-GE
Other Guidelines
Displays must be able to Frequency:
Display images in same display size Monthly
even if from different acquisition stations Objective:
with different pixel sizes Assure the monitor is calibrated
Annotate image information, image Brightness and contrast settings are at an
identification, & technical factor appropriate level for the reading of the images on
information the review workstation.
Display simultaneously a set of current &
prior conventional four-view screening Procedure:
mammogram images At the RWS the Start Calibration is selected.

As displays age, luminance and color temperature are apt to change. TOTOKU's Medivisor will keep your displays performing to DICOM standards. Medivisor is included in the calibration kits, along with photo se

Display Device Calibration Check-GE

The pattern is examined carefully for the following

features: As displays age, luminance are apt to
verify that the 0%-5% contrast is visible change. Monitor testing will keep your
verify that the 95%-100% contrast is visible displays performing to DICOM
verify that each gray level step from 0% to 100% can be standards. The calibration kit, along
distinguished from the adjacent squares. For example,
that you can distinguish the 0% square from the 10% with photo sensor or "puck" for
square , etc. luminance testing and adjustment is
verify that the line-pair images at the center and corners included.
of the SMPTE pattern are distinguishable.

Digital Image Presentation Issues

This test is no longer used with the new

LCD monitor screens. Time required to display an image
The only time the puck testing is needed on the WS 3 seconds or less
if facilities still use the CRT (Cathode Ray Displays should be able to
Tube) monitor. accommodate fast & easy navigation
Remember this test records luminance between old & new studies
levels on the monitor screen.

27
 1/9/2013

Digital Image Presentation Issues Digital Image Presentation Issues

Hanging protocols should be WS software tools must include

Flexible Window/level
Tailored to user preferences Zoom/pan
Specifically for mammography Specific recommendations regarding
with proper labeling & orientation types of tools to be used & how to
of images use them most effectively do not
exist

Digital Image Presentation Issues Monitor Secondary Display (TWS)

Technologists workstations used to judge image

quality during acquisition should be as similar as
WS should accommodate & possible to the RWS
display images from several Resolution (may have less)
modalities Maximum & minimum luminance
Contrast ratio
Ambient lighting
Conformance to DICOM
Zoom & pan

Monitor Clinician WS Repeat Analysis Check

Frequency:
Quarterly. For the repeat rate to be meaningful, an analysis
Used to review images as an adjunct to the period that yields a patient volume of at least 250 patients or
official interpretation by a radiologist 1,000 exposures is needed.
May not need as high resolution as RWS
Objective:
To determine the number and cause of repeated digital
mammograms. Analysis of this data can help identify ways to
improve system efficiency and reduce digital retakes and
patient exposure.

28
 1/9/2013

Repeat Analysis check

The applicable MQSA Quality Mammography Standard is: Action Limit:
900.12(e)(3)(ii)
Quarterly quality control tests. Facilities with screen-film systems shall If the total repeat rate changes from the rate
perform the following quality determined for the previous analysis period by more
control tests at least quarterly: than 2.0% of the total exposures included in the
(ii) Repeat analysis. If the total repeat or reject rate changes from the analysis, the reasons for the change must be
previously determined rate by
more than 2.0 percent of the total films included in the analysis, the determined. Any corrective actions taken must be
reason(s) for the change shall be determined. recorded and an assessment must be made of their
Any corrective actions shall be recorded and the results of these corrective effectiveness.
actions shall be assessed.

Compression Force Test Breast Compression

Frequency: Breast compression is equally important for digital

Initial installation and then every 6 months. mammography as it is for film screen. It contributes
to digital image quality by immobilizing the breast
(reduces motion unsharpness), producing a more
Objective: uniform, thinner tissue (lowers scatter radiation,
To assure that the mammographic system can more even penetration of x-rays, less magnification
provide adequate compression in power driven and or geometric blurring, less anatomical
manual modes and that the equipment does not superimposition), and lowering dose
allow too much compression to be applied.

Compression force test Printer

Procedure: Objective:
This test is not unique to digital mammography To ensure optimal quality of the film printer output,
systems. follow the QC developed by the manufacturer of the
Follow accepted mammographic QC procedures to device.
perform this test.
Record the results. If the printer is used with a film processor
Action Limit-GE &Lorad incorporating wet chemistry processing, follow the
The maximum compression force for the initial QC program developed by the manufacturer of the
power drive must be between 11 and 20 daN printer.
(25-45 lb.)

29
 1/9/2013

Hardcopy Printing - Considerations Hardcopy Printing - Considerations

FDA recommends only printers
specifically cleared for FFDM by the FDA requires the ability to print FFDM
FDAs ODE (Office of Device images of final interpretation quality
Evaluation) to film
MQSA does allow other printers to be Manufacturers guidelines should be
used followed
ACR strongly recommends only FDA -
cleared printers be used for digital
mammography

Hardcopy Printing - Considerations Hardcopy Printing - Considerations

FDA requires all printers used with an FFDM At present, no accreditation body reviews
unit softcopy images
Comply with a quality assurance program FDA recommends
that is substantially the same as that
Softcopy images be of such quality that if
recommended by the FFDM
they were submitted they would pass the
manufacturer
phantom & clinical image review process
That they pass the phantom & clinical of the facilitys accreditation body
image review process of the facilitys
accreditation body

QC testing for printers and monitors

without QC manuals

Procedure: In some cases the QC manual for the digital

Follow the manufacturers recommended quality mammography unit instructs the facility to test
control procedure. monitors and printers according to the components
Chart results. QC manual.
In these cases, it is the responsibility of the facility to
ensure that it obtains and follows the components
QC manual for its monitors and printers

30
 1/9/2013

Same printer or monitor with FFDM units

from different manufacturers DICOM Printer Quality Control

For facilities using FFDM units from For Lorad Selenias

different manufacturers, each with
its own QC requirements for printers
and monitors, there is some
uncertainty regarding the QC tests to
perform on these components.

DICOM Laser Printer Quality Control

Close any open examinations

Select test patterns from the Admin Menu
Select SMPTE from the Pattern drop-down menu
Select from the Output Size drop-down menu the
image size 18X24cm paddle
Uncheck the Print true size
Select the printer from the Output Devices drop-
down menu
Click Send to print the SMPTE pattern on the selected
printer
An image Queued dialog box appears informing you
that the image is being sent. Click on OK to close it.
Click close to exit the Test Pattern dialog

MD- Speed, DD- Contrast, LD- B&F, DC- Direct Current

31
 1/9/2013

Most printers are dry laser printers The film is exposed with a scanning laser. After
compared to wet chemistry printers. These exposure, the film is heated to a temperature of
devices were introduced in 1996. They 120 degrees for 24 seconds to process the image.
require specialty film which is After the image has been recorded, the film,
photothermographic that uses silver immediately after it is ejected from the printer, is
still in the processing image development.
behenate rather than silver halide to
produce the image and is processed Light from the viewbox illuminator can cause
slight changes in the optical density.
thermally rather than with liquid developer
and fixer.

Example 1 Example 2

Each FFDM manufacturer QC manual requires Each FFDM manufacturer QC manual requires that
that the same or equivalent test be done, but different but equivalent test be done.
In this case facilities may perform only one of the tests at
at different time frequencies. the more stringent frequency. The medical physicist
In this case facilities need to perform the should provide a written statement for the facilitys QC
records, indicating that in his or her opinion, the two tests
test at the more stringent frequency. are equivalent.

Example 3 FUJI CR QC
Baseline Tests-first measurement for each test
Each FFDM manufacturer QC manual requires Printer QC
that different test (not equivalent) be done. Monitor QC
CNR Weekly Check
In this case facilities need to perform each Image Quality
S Value Confirmation
test at the frequency required in the System Resolution
respective FFDM manufacturer QC manual. CR Reader Scanner performance
AEC system performance assessment
Imaging plate fog
System artifact evaluation
Dynamic range
Primary erasure
Inter-plate consistency
Dose

32
 1/9/2013

Image Plate Erasure FUJI CR QC

Using secondary erasure erase each plate first thing each morning
WEEKLY
CNR
Phantom Image
Printer QC
Monitor QC

FUJI CR QC Fuji CNR Check

Monthly Objective:
Visual Checklist To establish an operating level of Contrast-to-
Quarterly Noise Ratio (CNR) at a specific exposure and
Repeat Analysis weekly confirm that the CNR remains
Semi-annual consistent, within limits over time at the same
Compression exposure setting.
Imaging Plate Fog

4 cm acrylic on bucky with 0.2 mm thick

Establish a baseline CNR with 5 day average of the
aluminum object on top positioned as shown in
CNR valueswhen doing the 5 day average diagram.
remove and replace the aluminum between
Compression paddle in contact
exposures since positioning of it may have some Exposure: Mo/Mo 26 kVp @125
effect on the calculated CNR from week to week. mAs

Interval processing time-control time

Erase dedicated QC cassette using secondary interval between X-ray exposure and
reading of the IP. It must be
erasure consistent to Reduce the influence of
IP fading characteristics on the test.
5 or 10 minutes

33
 1/9/2013

Reestablish (5 day average) new baseline if: Performance Criteria CNR

X-ray tube replacement Must be within 20% of the baseline

Filter replacement image (CNR) established
Replacement of compression paddle
Change in phantom used
Change of IP/and or cassette used If failed must be corrected before any
Change of grid further examinations
Change of x-ray generator
Change of CR reader calibration

Fuji Phantom Image

Objective: Expose on technique used for 50/50

To assure that contrast, uniformity and image quality breast (this was auto filter-LoRad unit)
(due to the x-ray exposure system, image reader and
printer and workstation) are maintained at optimum Photocell in center of wax insert and in
levels. same place every time
Done hardcopy and softcopy
Same compression thickness every time
Send to printer once processed

Phantom Hardcopy Phantom Softcopy

Background density center of phantom no less Plot the S value from the exposure
than 1.20
on the S value range line on chart
Density difference in disc and adjacent to disk
and subtract Score subtracting for artifacts
Plot on chart and plot mAs
Score subtracting for artifacts

34
 1/9/2013

Performance Criteria
Score 4-3-3
If hard copy are used for final interpretation the OD
must be within .20 of the established OL and the
DD must be within .05
If softcopy images are used for final interpretation
the S value must not vary by greater than 20% (the
S value of the phantom confirms the exposure unit
output and the FCRm reader sensitivity setting)
If criteria is not met for either must be corrected
before any further examinations are performed.

Fuji Printer QC Upcoming Changes in QC for FFDM

Objective- to assure the printer used for final New BI-RADs and lexicon changes
interpretation is performing according to the
New ACR FFDM QC Control Manual
manufacturers specifications
New Digital Phantom for FFDM
Possibly a new Phantom for DBT

ACR FFDM QC Manual Project ACR FFDM QC Manual Project

Subcommittee Charge:
ACR Subcommittee on Quality Assurance Design ACR Accreditation Phantom for FFDM
Clinical Representatives Write QC Manual for ACR FFDM
MITA Representatives Mammography
ACR Representatives Accreditation Program
Information written by
.Et al. Eric Berns, PhD

35
 1/9/2013

ACR Digital QC Manual What will be New?

Structure of Manual: Tech Section
Radiologists Section Enhanced positioning and image quality section
Clinical Image Quality Section
New Test: Monitor QC for the Radiologist
Radiologic Technologists Section
Medical Physicists Section
New Test: Facility QC Review
Educational, Guidance, and Troubleshooting Section New Format: Corrective Action Log
Glossary New Documentation: Facility Equipment Inventory
References Instructions for Mobile Units
Index Eliminating calculations (Yet to be determined)

The ACR FFDM Phantom 24 X 30

Design Summary ACR Digital QC Manual

Differences from screen-film phantom Benefits of Phantom Design
Eliminate subtraction for artifacts Provides view of entire detector artifact evaluation
W/L optimized for test objects optimizes for artifact eval
Add Fail for artifacts
Finer gradations of test objects
Improve specific rules for scoring Test objects go to smaller sizes
Change pass/fail criteria from AGD measurement & limit same as SFM Meets MQSA
4,3,3 to 2,3,2 Provides single image/exposure for evaluation(s)
**But, objects are the same (effective) size as SFM Minimal training (~ 25,000 Techs currently trained)
Provides basis for monitor and laser printer QC
Phantom
ACR Physics Reviewers
Can see scores and artifacts on single submitted film (or image)
Do not need different WW/WL settings

36
 1/9/2013

CIRS Model 020 BR3D Mammography

Phantom
The CIRS Model 020 BR3D Mammography The phantom consists of a set of six (6) slabs made of
Phantom was designed to assess detectability heterogeneous breast equivalent material that
of various size lesions within a tissue exhibits characteristics of real breast tissue and
demonstrates how underlying targets can be
equivalent, complex, heterogeneous obscured by varying glandularity. Each slab contains
background. This phantom provides more two tissue equivalent materials mimicking 100%
realistic challenges for standard screen and adipose and gland tissues swirled together in a
FFDM mammography systems as well as approximate 50/50 ratio by weight. One of the slabs
Tomosynthesis and breast Computed contains an assortment of micro-calcifications, fibrils
Tomography. and masses.

Thats enough QC!!

37