Technique Optimization in

Digital Mammography
Nicole T. Ranger, M.Sc.

[email protected]

Carl E. Ravin Advanced Imaging Labs

Department of Radiology
Duke University Medical Center

Index
1. Introduction to Optimization in Mammography
2. Characteristics of S/F vs. Digital Mammography
3. Technique Optimization Methodology
4. What does this mean to me in the Clinic ?

Part 1
Introduction to
Technique Optimization in
Mammography
3

Technique

Acquisition Parameters
1. X-RAY BEAM SPECTRUM
- target material
- filter material and thickness
- tube kilovoltage (kVp)
2. EXPOSURE LEVEL
- beam current x time (mAs)
4

Optimization
Any process or procedure which ensures that
doses due to appropriate medical exposure for
radiological purposes are kept as low as
reasonably achievable (ALARA) consistent with
obtaining the required diagnostic information ..
IAEA-TECDOC-1447 May 2005
5

Implications
A mammography image of adequate to superior
image quality is NOT acquired using an optimal
technique if the dose to the patient was higher than
necessary to yield a diagnostic image.
Must find optimum
balance between
dose & image quality

DOSE

QUALITY

Optimization in Mammography
In mammography, the objective is to produce
images that provide maximum visualization of
breast anatomy and the signs of disease without
subjecting the patient to unnecessary radiation
- IAEA
7

Normal Breast Anatomy

1. Chest wall
2. Pectoralis muscles
3. Lobules
4. Nipple
5. Areola
6. Duct
7. Fatty tissue
8. Skin

www.wikipedia.com

Features of Breast Cancer

in Mammography Imaging
Increased Density (relative to prior exam)
Architectural Distortion
Micro-calcifications
Masses
9

Optimized Imaging of the Breast

1. DIAGNOSTIC OBJECTIVES

Detect and characterize microcalcification

cluster patterns and morphology
Visualize breast parenchyma and subtle
architectural distortions
Detect soft tissue masses and assess
shape, size, degree of local invasion
10

Optimized Imaging of the Breast

II. CHALLENGES
A.

NORMAL BREAST DENSITY VARIATIONS

- Age, Genetics, Menstrual cycle (premenopausal women)

ADIPOSE

GLANDULAR

Source: McGill University Department of Medicine Online Mammography Tutorial

For adequate SNR2 choice of radiographic technique

must be patient- and context-specific

11

Optimized Imaging of the Breast

II. CHALLENGES
B.

HIGH TISSUE CONTRAST RESOLUTION REQUIRED

Fibro-glandular and tumor

tissue have similar
attenuation properties
Optimize beam quality to
maximize differential
absorption and
Source: Bushberg, The Essential Physics of Medical Imaging
12

Use COMPRESSION
Craniocaudal Mediolateral
Oblique
(CC)

scatter
contrast

kVp

(MLO)
dose

mAs

Scatter Fraction:
Source: Basset LW, Imaging the Breast,
Cancer Medicine, 6th ed

Breast Thicknesses:
2 cm - 8 cm

0.8 1.0
uncompressed

0.4 0.5
compressed

Source: Bushberg, The Essential Physics of Medical Imaging

13

Optimized Imaging of the Breast

II. CHALLENGES
C.

NEED FOR DEVICE-SPECIFIC OPTIMIZATION

Different:

- Technology, i.e. Receptor Sensitivity

- Target/Filter options
- Exposure and other controls
- Image Processing

May require different imaging protocols & techniques

for different technologies & systems, i.e. SF versus CR,
CR versus DR, Direct DR versus Indirect DR, etc
14

Optimized Imaging of the Breast

III. IMAGING SYSTEM REQUIREMENTS

High Receptor Sensitivity

High Spatial Resolution

DOSE

High Contrast Resolution

Low Noise
Reproducibility
- Consistent image quality & exposure
15

Optimized Imaging of the Breast

IV. PRIMARY QUALITY FACTORS

RESOLUTION

Focal Spot Size & Blur

Magnification,
Pixel Pitch
Patient Motion Blur

CONTRAST

Tissue Density, Thickness

X-Ray Beam Quality (Target/Filter/kVP)

NOISE

Receptor Sensitivity
mAs / Exposure
16

Optimized Imaging of the Breast

V.

TECHNIQUE FACTORS

Target (Anode):

Molybdenum (Mo)
Rhodium (Rh)
Tungsten (W)

Filter:

Mo, Rh, Al, Ag

Tube Voltage:

kVp

Exposure:

mAs

CONTRAST
NOISE
17

Optimized Imaging of the Breast

VI. BREAST DOSIMETRY
Mean or Average Glandular Dose (MGD/AGD)
MGD = DgN x E
DgN: Exposure normalized
glandular dose
(abs. dose / unit exposure)
E:

Entrance surface exposure

~ 2.0 mGy (3.0 mGy ACR limit)

Source: Bushberg, The Essential Physics of Medical Imaging

18

Part 2
The Advantages of Digital
Mammography

19

Analog Imaging Chain

(Screen-Film)
Diagnostic
Exposure

Detection/
Capture
(Latent Image)

Chemical
Processing
(D-logE* + Chem.)

Output
Display
(Hardcopy*)

logE

Imaging
System

*same S/F
system used
@ 3 exp.
levels

20

Technique Optimization
Screen-Film
Optimization criteria based on film density
Typical Target/Filter: Mo/Mo or Mo/Rh
Dependent on film processor performance
Impacted by speed and latitude of film
Film is capture, display and storage medium
21

Digital Imaging Chain

(CR, DR)
Capture
(Sampling,
Quant.)

Detection/
Diagnostic
Capture
Exposure (Latent Image)

Output
Display
(Hardcopy
Processing
/Softcopy)

Output
Display
(Hardcopy*)

D, L
D

logE or E

Imaging
System

CV

*same
system used
@ 3 exp.
levels

22

Technique Optimization
Digital Mammography
Optimization criteria can be subjective, semiobjective or objective, e.g. based on a
computed figure-of-merit (FOM)
Prerequisite: Detector calibration & QC
Typical Target/Filter: W/Rh
Acquisition, processing and display can
and should be optimized separately
23

Latitude: S/F vs DR
SF

high

DR

SNR2

Adapted from: Mahesh M., Radiographics (2004)

low
24

DIGITAL MAMMOGRAPHY
CR SYSTEM COMPARISONS
CLASS TYPE
MODEL
/PHOSPHOR*

MANUFACTURER

PIXEL
PITCH*
(mm)

ACTIVE TARGET/
AREA FILTER
(cm2)

CR

PSP
BaSrFBrI:Eu

CR35-X
CR85-X

AGFA

0.050

18 x 24
24 x 30

CR

PSP
BaFBrI:Eu
PSP
BaFBr:Eu
PSP
BaFI:Eu

ASPIRE
CLEARVIEW

FUJIFILM

0.050

18 x 24
24 x 30
DIRECTVIEW CARESTREAM 0.0485 18 x 24
24 x 30
REGIUS 190 KONICA0.04375 18 x 24
MINOLTA
24 x 30

CR
CR

?
?

* Source: NHS Report 06047 Computed Radiography (CR) Systems for Mammography (2006)

25

DIGITAL MAMMOGRAPHY
DR SYSTEM COMPARISONS
CLASS TYPE

MODEL

VENDOR

PIXEL
PITCH
(mm)

ACTIVE
RECEPTOR
AREA (cm2)

TARGET/
FILTER

DR

Senoscan

FISCHER/
HOLOGIC

0.054
(0.027)

21 x 29

Mo/Mo
Mo/Rh
Rh/Rh
W/Al

DR

Indirect
CsI
Slot-scan
CCD
Direct aSe

Aspire HD*
(Amulet)

FUJIFILM

0.050

18 x 24
24 x 30

DR

Direct aSe

Selenia

LORAD/
HOLOGIC

0.070

25 x 29

Mo/Mo
Mo/Rh
Rh/Rh
Mo/Mo
Mo/Rh

DR

Indirect
aSi:CsI

Nuance
PLANMED
Nuance Excel

0.085

17 x 24
24 x 30

* NOT FDA APPROVED IN U.S.

W/Ag
W/Rh
26

DIGITAL MAMMOGRAPHY
DR SYSTEM COMPARISONS
CLASS TYPE

MODEL

VENDOR

PIXEL
PITCH
(mm)

ACTIVE
RECEPTOR
AREA (cm2)

TARGET
/FILTER

DR

Direct aSe

Novation DR
Novation S
Inspiration

SIEMENS

0.070
0.070
0.085

18 x 23
24 x 29
24 x 30

Mo/Mo
Mo/Rh
W/Rh

DR

Indirect
aSi:CsI
Flat Panel

2000D, DS
Essential
Senographe

GE

0.100
0.100
0.100

19.2 x 23
19.2 x 23
24 x 30.7

Mo/Mo
Mo/Rh
Rh/Rh

27

DIGITAL MAMMOGRAPHY
IMAGING PERFORMANCE COMPARISONS
FILM

CR

INDIRECT DR

DIRECT DR

As a class DR systems typically have lower resolution than screen/film

Source: A. Smith white paper

28

Why go Digital ?
DIGITAL

SCREEN-FILM

29

Advantages of Digital
Increased throughput

Image processing

Increased sensitivity

Advanced applications

Increased latitude

Decreased dose

Better contrast resolution

Indefinite archival

Contrast adjustment

Simultaneous access

But ..
Lower spatial resolution (generally)
30

Digital is the Future

As of 12/2009:
59% of MQSA certified facilities had 1 or more
FFDM units
60% of all MQSA certified units were FFDM
New mammography unit sales in US almost exclusively digital
Source: MQSA website

31

Part 3
Technique Optimization
Methodology

32

Technique Optimization
Methods
Spectral Simulation using Monte Carlo methods
Experimental Studies using Objective Criteria (Figures-of-Merit)
Clinically-Relevant Task Based Observer Studies
Prospective or Retrospective Clinical Studies
Subjective Evaluation of Image Quality Phantom Data
33

Example
Technique Optimization Protocol
- objective figure of merit (FOM)
- quotient of

2
SdNR

to MGD

- computed for masses and calcs

34

Technique Optimization Protocol

Physical Setup
Siemens Mammomat Novation DR
Mo/Mo vs W/Rh
kVp: 23, 25, 27, 29, 31, 33, 35

Ranger et al, Med Phys 37 (2010

35

Technique Optimization Protocol

FIGURE OF MERIT
2

SdNR
FOM =
MGD

Quality
Dose

SdNR2 : Signal Difference to Noise Ratio Squared

MGD: Mean Glandular Dose (computed from spectral
estimates using Spectra*)
* Boone et al, Radiol 213 (1999)

36

Technique Optimization Protocol

Characterizing the Beam Quality & Exposure
Measure free-in-air exposure at each
beam quality: target/filter & kVp
Extrapolate to phantom surface using
inverse square law
Measure HVL at each beam quality using
narrow beam geometry and calibrated ion
chamber fitted with a Mammo probe.

37

Technique Optimization Protocol

Computation of SdNR2

Heel Effect
With Inclusions

Without Inclusions

Si : mean signal in ROI overlying inclusion

Sb: mean signal in ROI in background (same location)
38

Technique Optimization Protocol

FOM Results
0% Gland

50% Gland

100% Gland

MASSES

CALCS

39

Technique Optimization Protocol

Computation of the FOMR and Dose Savings
Relative FOM (FOMR) gives the image
quality improvement at the new
technique (W/Rh) in comparison to the
optimized reference technique
(Mo/Mo ) for the equivalent
glandular dose

FOM EVAL
FOM =
FOM REF
R

RELATIVE
DOSE
SAVINGS

(FOM )

R 1

40

Technique Optimization Protocol

Relative Dose Savings
Siemens Novation DR

41

Dose savings likely

to be much more
conservative .
WHY?
42

Breast Density:
What is typical or average ?
Common misconception that average breast density is
50% adipose tissue to 50% fibroglandular or 50-50

Ref: Yaffe et al, The Myth of the 50-50 Breast

43

Normal Density Variations

ACR BI-RADS BREAST DENSITY CLASSIFICATION SCHEME
Category 1

Category 2

Category 3

ADIPOSE

Source: McGill University Department of Medicine Online Mammography Tutorial

Category 4
GLANDULAR

44

Part 4
What does all this mean
in the clinic ?
45

Summary Key Concepts

Image quality must be optimized in the context of the
COST (RISK) associated with the patients
EXPOSURE (DOSE) and that is accomplished via
Technique Optimization
- Tradeoff between QUALITY and DOSE
- Optimization must be patient- and applicationspecific, i.e. tailored to patient size, exam type
and other relevant factors
46

Recommendations
1.

Automatic exposure (and other controls, e.g. automatic

kVp, Target-Filter selection, etc..) controls should be
verified as part of routine acceptance testing and checked
on a routine basis.

2.

AEC cells in conventional mammo units are located

beneath the grid. CR cassette structure/sensitivity varies
from vendor to vendor. Must calibrate the AEC for the
intended CR casette type and only use those cassettes

3.

Newer FFDM systems employ a virtual AEC in conjunction

with a short prescan to determine the exposure termination
conditions. Make sure it works as intended
47

Recommendations
4.

An imaging QA program must include monitoring of

the system in the context of its intended clinical use
a PROPERLY OPERATING imaging system is
NOT the same as one OPERATED PROPERLY

- Validate vendor-recommended techniques/protocols

- Establish written imaging protocols and SOPs
- Mandate ongoing staff training and in-services
- Conduct routine audits to ensure compliance
48

ACKNOWLEDGMENTS
Information or Assistance Provided by:
DUKE
Joseph Y. Lo, PhD
Ehsan Samei, PhD
Jay A. Baker, MD
Anne Jarvis

OTHER
Eric Gingold, PhD

SIEMENS
Thomas M. Mertelmeier, PhD

49

SAM Question #1
Which of the following is NOT an advantage of
digital mammography over screen film
mammography:
20%
20%
20%
20%
20%

a)
b)
c)
d)
e)

improved throughput
improved latitude
higher spatial resolution
decreased dose for comparable image quality
image processing and digital archival
50

10
50

SAM Question #1
Which of the following is NOT an advantage of digital
mammography over screen film mammography:
(a) improved throughput
(b) improved latitude
(c) higher spatial resolution
(d) decreased dose for comparable image quality
(e) image processing and digital archival
Answer: (c)

51

Ref: Williams, M.B., Fajardo L.L., Digital Mammography: Performance Considerations

and Current Detector Designs, Acad Radiol 3:429-437 (1996)
51

SAM Question #2
The technique factor that has the strongest impact on
digital mammography image quality as reflected in a
Figure-of-Merit (FOM) computed from the ratio of CNR2
or SdNR2 to MGD is:
20%
20%
20%
20%
20%

a)
b)
c)
d)
e)

focal spot size

mAs
field size
kVp
target / filter combination

52

10
52

SAM Question #2
The technique factor that has the strongest impact on digital
mammography image quality as reflected in a Figure-of-Merit
(FOM) computed from the ratio of SNR2 or SdNR2 to MGD is:
(a) focal spot size
(b) mAs
(c) field size
(d) kVp
(e) target / filter combination
Answer: (e)

53

Ref1: Optimization of Exposure Factors in Full Field Digital Mammography, Williams et al,
Med. Phys. 35: 2414-23 (2009)
Ref2: A Technique Optimization Protocol and the Potential for Dose Reduction in Digital
Mammography, Ranger NT, Lo JY, Samei E, Med. Phys. 37: 962-9 (2010)

53

SAM Question #3
In digital mammography, the approximate Mean
Glandular Dose (MGD) for a 5 cm thick average density
breast imaged using automatic exposure control with a
W/Rh target/filter combination would be closest to:
20%

20%

20%

20%

20%

a)
b)
c)
d)
e)

0.02 mGy
0.2 mGy
2.0 mGy
20.0 mGy
200.0 mGy

54

10
54

SAM Question #3
In digital mammography, the approximate Mean Glandular
Dose (MGD) for a 5 cm thick average density breast imaged
using automatic exposure control with a W/Rh target/filter
combination would be closest to:
(a) 0.02 mGy
(b) 0.2 mGy
(c) 2.0 mGy
(d) 20.0 mGy
(e) 200.0 mGy

55

Answer: (c)
Ref: Comparison of Acquisition Parameters and Breast Dose in Digital Mammography and
Screen-film Mammography in the American College of Radiology Imaging Network Digital
Mammographic Imaging Screening Trial, Hendrick RE, Pisano E, Averbukh A, Moran C,
Berns EA, Yaffe MJ, Herman B, Acharyya S, Gatsonis C, AJR 194: 362-369 (2010)

55

SAM Question #4
For a 4 cm breast with a composition ratio of 50%
glandular to 50% adipose, the magnitude of dose
reduction achievable for comparable image quality in
the transition from screen/film to digital mammography
is approximately:
20%
20%
20%
20%
20%

a)
b)
c)
d)
e)

10 %
20 %
40 %
60 %
80 %

56

10
56

SAM Question #4
For a 4 cm breast with a composition ratio of 50% glandular
to 50% adipose, the magnitude of dose reduction
achievable for comparable image quality in the transition
from screen/film to digital mammography is approximately:
(a) 10 %
(b) 20 %
(c) 40 %
(d) 60 %
(e) 80 %

57

Answer: (c)
Ref: A Technique Optimization Protocol and the Potential for Dose Reduction in Digital
Mammography, Ranger NT, Lo JY, Samei E, Med. Phys. 37: 962-9 (2010)

57

Thank you for your attention.

Email: [email protected]

OLD SLIDES

59

60

61

Source: IMS Giotto 3D brochure

62

63

Characterizing Breast Masses

Mass Shape

Mass Margins

64

Characterizing Microcalcifications
Low Probability or Typically Benign
Skin calcifications, lucent, polygonal shape
Vascular calcifications
Coarse, larger calcifications

Intermediate Probability

NEED EXAMPLE
IMAGES/FIGURES

Indistinct or amorphous microcalcifications

High Probability
Clusters of heterogeneous or pleomorphic calcifications,
irregular size & shape & < 0.5 mm diameter are suspicious
Fine linear or branching calcifications < 1 mm in with are
associated with necrotic cancer cells
65

Need a image quality- & dosesensitive metric to objectively

assess optimization

66

Typical Calibrations (DR)

Correction for bad detector elements (dels)
Flat-field correction
Correction for radiation field non-uniformity (i.e.Heel Effect)
Gain and offset correction for each del
Correction for velocity variation during scan (CR)

Correction for geometric distortion *

* Primarily in CCD lens-based imaging systems

67

study published in September 2005 in the New England Journal of

Medicine compareddigital mammograms to film mammograms. The
study involved 49,000 women in North America with no known signs of
breast cancer. The women were screened using both digital and film
mammograms at the beginning of the study and again one year later.
Breast cancer was found in 335 of the women. The researchers
determined that digital mammograms were superior to film
mammograms for three groups:women under 50 years of agewomen
with dense breastswomen who have not yet gone through menopause,
or who have been in menopause less than one year.Digital
mammograms did not prove to be more beneficial for post-menopausal
women over age 50 that do not have dense breasts. Additionally, both
forms of mammogram had the same rate of false positives.
68

Recommendations
Take advantage of the free resources available
Invest in a high quality IEC-compliant edge device
Use lower purity legacy Al not the 99.9% purity Al
specified by the standard (test Al before use)
Review the fundamentals of DQE physics
Study the body of literature on DQE testing
Obtain the IEC standards for reference
Recruit a DQE Mentor
69

Thank you for your attention.

Email: [email protected]

EXTRA STUFF

71

72

ACKNOWLEDGMENTS
Information &/or Assistance Provided by:
DUKE
Joseph Y. Lo, PhD
Ehsan Samei, PhD
Jay A. Baker, MD
Anne Jarvis
SIEMENS
Thomas M. Mertelmeier, PhD

73

Introduction
CONTEXT

1 in 8 women (13%) will be diagnosed with breast

cancer in their lifetime
In 2006, 191,410 new breast cancer cases
In 20__ , __ screening mammography exams were
conducted in the US and __% were performed using a
digital mammography imaging system
74

Characteristic Photons:
Moly: 17.9, 19.5
Rhodium: 20.2, 22.7
Best spectra with kVp set 5-10 kVp above K-edge

75

3.0

Mo/Rh

2.5

DOSE
(mGy)

2.0
1.5

Mo/Mo

1.0

W/Rh

0.5
0.0
2

BREAST THICKNESS (CM)

Source: IMS Giotto 3D brochure

76

Radiographic Characteristics
of Breast Tissue
Tissue Type Density Atomic #
Adipose
Glandular
Calcification

77

Breast Imaging Statistics

Breast cancer #__ cause of death in women aged XX-YY
Lifetime risk: 1 in 8 women will be diagnosed
In 2006, ____ new cases of breast cancer were diagnosed; approx.
half of which were diagnosed as a result of screening
mammography
Risk of inducing a breast cancer as a result of exposures
associated with lifetime of screening mammography: ____
In 20__ , ____ screening mammography exams were conducted in
the US and ___% of those studies were performed using a digital
mammography imaging system
Sources: http://apps.nccd.cdc.gov/uscs/

78

Technique Optimization Protocol

Characterizing the Beam Quality & Exposure
Measure free-in-air exposure at each
beam quality: target/filter & kVp
Extrapolate to phantom surface using
inverse square law
Measure HVL at each beam quality
using narrow beam geometry and
calibrated ion chamber fitted with a
Mammo probe
79

GE Senographe
CR: PSP

Indirect DR: Cs(I) TFT

Hologic Selenia
Direct DR: aSe TFT

80

GE Senographe
CR: PSP

Indirect DR: Cs(I) TFT

INDIRECT DETECTION

Hologic Selenia
Direct DR: aSe TFT

DIRECT DETECTION

81

Computed Radiography

Source: Basset LW, Imaging the Breast, Cancer Medicine, 6th ed

82

Indirect DR
Cs(I) TFT Array

Source: Basset LW, Imaging the Breast, Cancer Medicine, 6th ed

83

Direct DR
aSi TFT Array

Source: Basset LW, Imaging the Breast, Cancer Medicine, 6th ed

84