advances)in)digital) mammography) - bem vindo ao sítio … 3/auditorium/2 advances in...

Advances in digital mammography

Gisella Gennaro Veneto Ins7tute of Oncology (IOV), IRCCS

Padua -‐ Italy

Lisbon -‐ Sep 20, 2013

Outline

1.  Digital breast tomosynthesis (DBT)

2.  (Contrast-‐enhanced) spectral mammography

3.  Quan7ta7ve breast density

G. Gennaro – Lisbon -‐ Sep 20, 2013

Limita7on of standard mammography

Breast is 3D

X-‐rays (invisible)

TransmiPed X-‐ray image (invisible)

DETECTOR

MAMMOGRAM (visible)

Signal propor7onal to X-‐ray photon intensity/number

Mammography image is 2D

(1/3)



Limita7on of standard mammography ü  Everything between the beam entrance and the breast exit is projected onto a

plane ü  A mammography image is an absorp7on map produced by superimposed

structures ü  Breasts are different in structure and produce very different absorp7on maps

(2/3)


Limita7on of standard mammography LESION DETECTION IS BASICALLY A PROBLEM OF LOCAL CONTRAST

(3/3)


What tomosynthesis is

ü  The digital applica7on of old techniques called “laminography”

ü  A “small angle CT”, where mul7ple planes of an anatomical part are reconstructed ONLY IN ONE SPATIAL DIRECTION , star7ng from a limited number of low-‐dose, projec7on images, acquired within a limited angular aperture


The idea is not new… A Bocage (1917): developed the «Biotome», a radiographic technique which allowed «cueng» the human body in longitudinal slices and, at the end, to select the level needed (Great War – to determine the depth of a metal foreign body in the body of a pa7ent). Never produced.

Vallebona 1930 –stra@graphy Ziedses Des Plantes 1931 – planigraphy

Fig. 5 Fig. 6

Chaoul & Grossman 1934 – tomography

Sans & Porcher 1947– politome (mul@direc@onal tomograph)


Digital detectors made it possible

RX digital detector

pre-processing post-‐processing

print

display dose

control

ACQUISITION DISPLAYING



How tomosynthesis works

Radiology 1997, 205: 399-‐406

ü  Senographe DMR modified ü  Digital detector developed by GE ü  ACR phantom ü  4 mastectomy specimens

CONCLUSION: «Digital mammographic system make breast tomosynthesis possible. Tomosynthesis may improve the specificity of mammography with improved lesion margin visibility and may improve early breast cancer detec@on, expecially in women with radiographycally dense breasts.»

Breast tomosynthesis feasibility



Breast tomosynthesis «evolu7on»

Equipment Med Phys 2013, 40

ü  S ame t e chn i c a l s p e c i fi c a7on s a s t h e «mammography plakorm» on which the tomosynthesis system has been developed (same detector, same X-‐ray source)

ü  Different acquisi7on protocols (tube mo7on, angular range, number of projec7ons)

ü  Different reconstruc7on algorithms ü  Without an7scaPer grid

NOWADAYS, TOMOSYNTHESIS HAS BEEN CONSIDERED BY MANUFACTURERS MORE INTERESTING THAN OTHER 3D TECHNIQUES (like breast CT) BECAUSE IT IS NOT A SEPARATE MODALITY, BUT IS JUST A «MODIFIED MAMMOGRAPHY SYSTEM». IT CAN BE SOLD AS AN OPTIONAL OF THE MAMMOGRAPHIC UNIT


Tomosynthesis and radia7on dose

ü  Most of the studies with prototype devices kept «wide margins» with radia7on dose, by applying the «non-‐inferiority principle» versus other well accepted technologies, like screen-‐film mammography. Dose reduced moving to commercial systems.

ü  It is difficult to go in depth in tomosynthesis dosimetry, the clinical protocol with DBT being s@ll under discussion: DBT alone or added to mammography? One or two views?…). USUALLY, TOTAL DOSE WITH DBT SHOULD NOT EXCEED THE TOTAL DOSE DELIVERED FOR MAMMOGRAPHY.

ü  The dose parameter used is s7ll the average/mean glandular dose (AGD o MGD)

Sechopoulos et al. (2007), Med Phys 34:221

Rela7ve Glandular Dose

Thickness crossed by the perpendicular beam

GIVEN THE SAME ENTRANCE DOSE (PER P R O J E C T I O N ) , D O S E A C T U A L L Y CONTRIBUTING TO PRODUCE THE IMAGE CHANGES WITH THE ANGLE


DBT systems

Image detector / X-‐ray source

Manufacturer GEH Hologic IMS Philips Planmed Siemens

System Senographe Essen7al

Selenia Dimensions GioPo TOMO MicroDose Nuance

Excel DBT Mammomat Inspira7on

Detector FPD CsI FPD a-‐Se FPD a-‐Se Photon coun7ng FPD a-‐Se FPD a-‐Se

Pixel size (μm) 100 70 MG 140 DBT 85 50 85 85

Anode Mo e Rh W W W W Mo e W

Filtra7on 0.03 mm Mo 0.025 mm Rh

0.05 mm Rh MG 0.7 mm Al DBT

0.05 mm Rh 0.05 mm Ag 0.5 mm Al 0.06 mm Rh

0.075 mm Ag 0.03 mm Mo 0.05 mm Rh

(1/2)


Tomosynthesis Manufacturer GEH Hologic IMS Philips Planmed Siemens




Detector mo@on sta7c rota7ng sta7c con7nuous slit scan rota7ng sta7c

Tube mo@on step-‐and-‐shoot con7nuous step-‐and-‐shoot con7nuous con7nuous con7nuous

Angular range (°) 25 15 40 11 30 50

N° of projec@ons 9 14 13 21 15 25

Scan @me(s) 7 4 12 3-‐10 20 25

mAs/projec@on variable* uniform variable* uniform uniform uniform

Reconstruc@on algorithm itera7ve FBP itera7ve itera7ve itera7ve FBP

Development stage** CE mark FDA approval CE mark prototype prototype CE mark

*different dose distribuIon between central projecIon and angled projecIons ** possible evoluIon

(2/2)


DBT systems

DBT: angle and number of projec7ons

Manufacturer GEH Hologic IMS Philips Planmed Siemens




Angular aperture (°) 25 15 40 11 30 50

N° of projec7ons 9 14 13 21 15 25

by Siemens G. Gennaro – Lisbon -‐ Sep 20, 2013

ü  Wide differences in angular aperture across manufacturers ü  The number of projec7ons is limited by the overall radia7on dose (and by the

angle itself) ü  Small angles show advantages to depict small objects (ex. microcalcs) ü  Wide angles show advantages to depict «large» objects (ex. masses)

DBT: projec7ons vs. tomographic planes

proj #2 proj #3 proj #4 proj #5

proj #6 proj #7 proj #8 proj #9 proj #10

proj #11 proj #12 proj #13 proj #14 proj #15

© V

enet

o In

stitu

te o

f Onc

olog

y, P

adua

, Ita

ly

proj #1

RECONSTRUCTION ALGORITHM

IN-‐FOCUS PLANE

IN-‐FOCUS PLANE


ü  DBT «slices» have no thickness, they are planes

ü  The value reported as «slice thickness» (typ. 0.5-‐1.0 mm) is the sampling interval

ü  Breast volume reconstruc7on possible only in the direc7on of breast compression

Med. Phys. 2004, 3:2636-2647

“DIRECT” ALGORITHM •  back-‐projec7on (BP) •  filtered back-‐projec7on (FBP)

ITERATIVE ALGORITHMS •  Algebraic methods: -‐ Algebraic Reconstruc7on Technique (ART) -‐ Simultaneous Algebraic Reconstruc7on Technique (SART) -‐ Simultaneous Itera7ve Reconstruc7on Technique (SIRT)

•  Sta7s7cal methods: -‐ Maximum Likelihood (ML)

TOMOSYNTHESIS ACQUISITION GEOMETRY IS LIMITED FOR THE FOLLOWING RECONSTRUCTION: •  LIMITED NUMBER OF PROJECTIONS •  LOW DOE PROJECTIONS •  LIMITED ANGULAR RANGE

Reconstruc7on algorithms


FBP SART

DBT: strengths

ü  TOMOSYNTHESIS IS VERY EFFECTIVE IN DEPICTION OF ARCHITECTURAL DISTORSIONS AND SPICULATED MASSES

ü  FALSE POSITIVES FROM SCARS / PREVIOUS BIOPSIES …

ü  VERY GOOD FOR MASSES ü  ALLOWS SEPARATION OF MASSES IN-‐FOCUS IN DIFFERENT

PLANES


© Veneto Institute of Oncology, Padua, Italy

DBT: poten7al weaknesses ü  MICROCALCIFICATIONS ARE DISTRIBUTED IN DIFFERENT PLANES AND CAN LOOSE

CLUSTER APPARENCE ü  THEY LOOK LIKE «BRIGHTER» THAN IN STANDARD MAMMOGRAPHY, AND THIS MIGHT

INDUCE THE READERS TO UNDERESTIMATE THE PROBABILITY OF MALIGNANCY, CALCIFICATION DENSITY BEING USUALLY ASSOCIATED TO BENIGN LESIONS

G Gennaro et al (20113), Eur Radiol DOI 10.1007/s00330-‐013-‐2831-‐0

«Clarity»

ML Spangler et al (2011), AJR 196:320-‐324


Slab computa7on •  Slabs : -‐ combine informa7on spread on several consecu7ve slices (microcalcs)

-‐ overlap (half thickness): reduce risk of lesions spread on two consecu7ve slabs -‐ reduced dataset (12 images as an average to scroll the en7re volume)

Courtesy of Sylvain Bernard, GEH


thickness

overlap

z

MX

DBT: slab MX MLO

© Veneto Institute of O

ncology, Padua, Italy

Clinical results

ü Many published results have been obtained by prototype devices

ü  (SOME «PREFERENCE» STUDIES) ü MOSTLY RETROSPECTIVE STUDIES ü  FIRST RESULTS FROM PROSPECTIVE STUDIES

CARE NECESSARY TO COMPARE CLINICAL STUDIES WITH DIFFERENT DESIGNS (IS WRONG TO CONSIDER ONLY CONCLUSIONS)


ü  Images obtained with the two compared modali7es, separately evaluated, taking all the ac7ons to limit possible biases reduc7on; usually, readers are asked to localize features and rate them according to given scales.

ü  Typical scales: probability of malignancy or BI-‐RADS score (5-‐7 steps) ü  Truth is established from histology (in case of surgery or biopsy) and/or long-‐

term follow-‐up (at least 1 y)

ü  Radiologists ra7ngs are compared with the truth, and differences between clinical performance variables (Se, Sp, PPV, PNV, ROC, AUC, FROC, etc) for each modality analyzed

FN / TN FP / TP

1 2 3 4A 4B 4C 5

BI-RADS scale

Clinical performance studies


(1/2)

1° author/journal Popula@on Readers Comparison Results Andersson et al / Eur Radiol 2008 36 cases (40 K) 2 DBTMLO vs. MX Cancer visibility superior with

DBT

Gur et al / AJR 2009 125cases (35 K) 8 DBT + MX vs. MX 30% reduc7on recall rate

Spangler et al / AJR 2012 100 cases (20 K) 5 DBT vs. MX Calcs: slightly bePer detec7on with DBT

Teerstra et al / Eur Radiol 2010

513 cases (112 K) 1 DBT vs. MX

ü Same Se: 92.9% ü Sp: MX = 86.1%; DBT = 84.4%

Gennaro et al / Eur Radiol 2010 376 cases (63 K) 6 DBTMLO vs. MX AUC: DBT non-‐inferiority

Gur et al / AJR 2011 125 cases (35 K) 8 DBT + MX vs. MX DBT + MX superior in lesion detec7on

Wallis et al / Radiology 2012 130 cases (40 K) 10 DBTMLO vs. MX DBT vs. MX

ü DBT non-‐inferiority ü DBT superiority

Svahn et al / BJR 2012 185 cases (95 K) 5 DBTMLO vs. MX DBT bePer performance (JAFROC)

Rafferty et al / Radiology 2012 312 cases (48 K) 312 cases (51 K)

12 15

DBT + MX vs. MX DBT+MX > accuracy (AUC) DBT+MX < recall rate

Zuley et al / Radiology 2012 217 cases (72 K)

8 DBT+MX vs. MX + extra views

Superiority DBT+MX

Gennaro et al / Eur Radiol 2013 469 cases (68 K)

6 DBTMLO + MXCC vs. MX per-‐breast

DBTMLO + MXCC non-‐inferiority Superiority for Sp benign

Gennaro et al / Eur Radiol 2013 463 cases (77 K)

6 DBTMLO + MXCC vs. MX per-‐lesion

Superiority lesion detec7on and characteriza7on

Clinical performance studies


(2/2)

Prospec7ve trials Oslo trial

•  17,960 women •  2-‐view DBT + 2-‐view MX vs. 2-‐view MX

Malmo trial

ü  higher detec@on rate (27% increase) ü  lower recall rate (15%) ü  Higher detec@on invasive cancers (40%)

•  15,000women •  1-‐view DBT + 2-‐view MX vs. 2-‐view MX


•  7292women •  2-‐view DBT + 2-‐view MX vs. 2-‐view MX

ü  higher detec@on rate ü  lower recall rate

•  7000 women •  2-‐view DBT vs. 2-‐view MX in screening recalled cases

STORM trial (Italy) Lancet Oncol 2013:14

TOMMY trial (UK)

ü  Clinical trial are 7me consuming and complex, and it is forbidden to change the study design while it is going on.

ü  This is the major reason why trial designs are «conserva7ve» (in par7cular, prospec7ve studies).

ü  Radiologists are not «trauma7zed» from DBT evalua7on, images being very similar to mammography images; however, it might be necessary some 7me to learn (for example to properly interpret some DBT features)

ü  While DBT is allowed to be used together with mammography, some risks of «misuse» exist.

Will tomosynthesis replace mammography ?


ü  Which protocol ? DBT alone? How many projec7ons (CC/MLO/both)? DBT combined with mammography? Which combina7on ?

ü  How evalua7ng DBT ? Plane-‐by-‐plane ? Cine-‐loop ? Headache? ü  Review 7me: is it «compa7ble» with the screening workflow in real

life ? ü  Is DBT sufficient for dense breasts or do we s7ll need ultrasound ? ü  Cost-‐effec7veness MX+DBT vs. MX+US ü  Layout to compare possible combined protocols with prior

examina7ons. ü  Image storage (1-‐3 GB vs. 200 MB per pa7ent) and retrieving from

PACS.

Open ques7ons


1.  DBT is a quasi-‐3D technique which reconstructs breast volume star7ng from a limited number of low-‐dose projec7on within a small angle. Differences across systems.

2.  Strong with masses and architectural distorsions, s7ll ques7ons on detec7on of microcalcs (CAD?)

3.  Several retrospec7ve studies, different in designs and popula7ons: combina7on of DBT with MX improves clinical performance (different combined protocols).

4.  Screening perspec7ve studies: significant increase in cancer detec7on rate and decrease in recall rate.

5.  Some ques7ons s7ll open

Take home messages


Outline





Limita7on of standard mammography

Breast is 3D


TransmiPed X-‐ray image (invisible)

DETECTOR

MAMMOGRAM (visible)

Signal propor7onal to X-‐ray photon intensity/number

Mammography image is 2D



Subtrac7on techniques in mammography

DIGITAL RADIOGRAPHY

ENERGY DEPTH

TIME SPACE

ü  TO ENHANCE CONTRAST OF FINDINGS

ü  TO INCREASE LESION CONSPICUITY

temporal subtracIon (angiography, breast)

3D techniques (CT, tomosynthesis)

phase contrast

Combina9on of two images/sets of images obtained with different spectra


Figure 2. Temporal contrast-‐enhanced mammography images. (a) shows a craniocaudal digi7zed screen-‐film mammogram obtained in a pa7ent with infiltra7ng ductal carcinoma. (b) shows a craniocaudal contrast-‐enhanced digital subtrac7on image obtained 1 minute a�er the start of contrast medium injec7on and shows a small nodule with rim enhancement of the en7re mass (arrow). (c) shows subtrac7on image obtained 10 minutes a�er start of contrast medium injec7on and shows washout of contrast medium. Note that only a single breast can be obtained in one view and that in this specific case, the breast needed to be compressed for 10 minutes. Images used with permission (Jong et al., Radiology 2003; 228: pages 842-‐850).

Temporal subtrac7on


Lobbes et al., Clin Radiol 68:935 (2013)

Dual-‐energy techniques ü  TWO IMAGES ARE OBTAINED USING TWO DIFFERENT SPECTRA, WELL

SEPARATED IN ENERGY, USUALLY INDICATED WITH «LOW ENERGY» AND «HIGH ENERGY». FOR THIS REASON, DUAL ENERGY IMAGING IS OFTEN REPORTED AS «SPECTRAL IMAGING».

ü  HYBRID IMAGES (ARE PRODUCED BY NON-‐LINEAR COMBINATION OF THE LOW-‐ AND THE HIGH-‐ENERGY IMAGES

DUAL ENERGY TECHNIQUES ARE USED: •  IN COMPUTED TOMOGRAPHY (DECT) •  IN DIGITAL RADIOGRAPHY (chest, musculoskeletal, etc) •  TO DIFFERENTIATE MATERIALS LIKE IN DENSITOMETRY (DEXA, dual-‐energy x-‐ray

absorp7on) •  IN DIGITAL MAMMOGRAPHY

CONTRAST MEDIA ARE USED IN DUAL ENERGY MAMMOGRAPHY


Contrast-‐enhanced spectral mammography

∼ 20s

ü  Mo or Rh target ü  Mo or Rh filter ü  26-‐31 kVp

ü  Mo target ü  0.3 mm Cu + 0.3 mm Al ü  45-‐49 kVp

Iodine injec@on

LOW-‐ENERGY (below iodine k-‐edge)

HIGH-‐ENERGY (above iodine k-‐edge)

RADIATION DOSE: Low+High energy images: ~20% extra dose than a standard mammogram

DUAL-‐ENERGY SUBTRACTION IMAGE

Iodinated CM: same dose as used in CT (1.5 ml/kg)


Dromain et al., Eur J Radiol 69:34 (2009)

(1/3)

Lobbes et al., Clin Radiol 68:935 (2013)

ArvaniIs et al., PMB 54:6041 (2009)

TWO SPECTRA, SEPARATED IN ENERGY


Iodine k-‐edge


(2/3)

Low-‐energy image: equivalent to standard mammogram

Subtracted image



(3/3)

Spectral mammography


Clinical results


110 cases (148 lesions) 6 readers CESM+MX+US vs. MX+US

52 pa7ents Known cancer Pre-‐surgical CESM vs. MRI

1.  Subtrac7on techniques can be applied to mammography, either temporal or dual-‐energy subtrac7on. DE currently preferred.

2.  With contrast media (typ. Iodine) 3.  Promising as «problem solver» for cases referred to breast

MRI. Faster and easier than MRI.

Take home messages

Outline





What breast density is

PC Johns, MJ Yaffe, «X-‐ray characterizaIon of normal and neoplasIc breast Issue», Phys Med Biol 1987, 32:675-‐695

fat

fibroglandular 7ssue

Fat has lower aPenua7on coefficient than fibroglandular 7ssue, which means it is more transparent to X-‐rays. Usually, in mammography images, fat is dark, while fibroglandular 7ssue is brighter. The denser the breast, the larger the bright parts in the image.


Why «measuring breast density ?

fat


1 To «officially state» poten7al difficults in diagnosis (radiologists are well aware about differences in lesion detectability depending on breast paPern/density)

2 To «quan7fy» a risk factor

«…substan7al literature shows that more extensive density is associated with an increased risk of breast cancer. Women with dense 7ssue in 75% or more of the breast have a risk of breast cancer 4-‐6 7mes as great as the risk among women with liPle or no dense 7ssue.»

ALL PUBLISHED PAPERS ON BREAST DENSITY, QUOTE THE SENTENCE «WOMEN WITH DENSE BREASTS HAVE A CANCER RISK INCREASED BY 4-‐6 TIMES COMPARED TO WOMEN WITH FATTY BREASTS»


fat


Breast density and risk factors


Nature 2012, vol 490

Human-‐based evalua7on of breast density

Wolfe’s Classifica@on (1976)

N1 (N=normal)

the breast consists mainly of fat, lower risk for breast cancer

P1 (P=prominent ducts)

this paPern includes fat as well as linear densi7es (enlarged ducts) occupying no more than 25% of the breast

P2 linear densi7es (from enlarged ducts) occupying more than 25% of the breast. They are prominently in the upper outer quadrant but may be distributed throughout the breast

Dy (Dy=dysplasia)

dense, radiopaque breast, highest risk for breast cancer

Categoria Boyd (1980) Tabar (1997) ACR BIRADS

1 A: 0% I : balanced propor7on of all components of breast 7ssue with a slight predominance of fibrous 7ssue

Type 1: extremely fat

2 B: 1-‐10% II : predominance of fat 7ssue Type 2: minimal density

3 C: 11-‐25% III : predominance of fat 7ssue with retroareolar residual fibrous 7ssue

Type 3: heterogeneous density

4 D: 26-‐50% IV : predominantly nodular densi7es Type 4: extremely dense

5 E: 51-‐75% V : predominantly fibrous 7ssue /

6 F: >75% / /

ü  4-‐6 categories ü  Defined by «descrip@ons» ü  Auempt to quan@fy «at a

glance» the «black and white» percentages in the image

*

* FILM MAMMOGRAMS IN THE 70s-‐80s LOOKED MUCH DIFFERENT THAN THE CURRENT IMAGES IN TERMS OF CONTRAST


BIRADS classifica7on of breast density

ALMOST ENTIRELY FATTY

SCATTERED FIBROGLANDULAR

HETEROGENEOUSLY DENSE

> 75% < 25% 25-‐50% 50-‐75%

EXTREMELY DENSE

ü  Categorical scales (descrip7ve) can be subjec7vely interpreted ü  Some breast paPerns are border line between two adjacent categories (1/2, 2/3, 3/4)

ü  APempt (unsuccessful) to «quan7fy» the percentage of glandular component


Padua, IOV Ø  2103 cases, 6 radiologists Ø  Moderate agreement (k=0.56)

Kappa value Agreement

< 0.2 Poor

0.21 – 0.40 Fair

0.41 – 0.60 Moderate

0.61 – 0.80 Strong

> 0.80 Near complete

0

200

400

600

800

1000

Obs 6

Obs 5

Obs 4

Obs 3

Obs 2

Obs 1 BIRADS 1 BIRADS 2 BIRADS 3 BIRADS 4


Inter-‐reader variability

Breast density can be quan7fied


Methods based on area

JW Bing et al. (1998), Analysis of mammographic density and breast cancer risk from digiIzed mammograms, Radiographics 18:1587

L-‐JW Lu et al. (2007), CompuIng mammographic density from a mulIple regression model constructed with image acquisiIon parameters from a full-‐field digital mammographic unit, Phys Med Biol 52:4905

ü  Algorithms: segmenta7on/texture/fractals, automa7c or semi-‐automa7c ü  PD = «dense» area/breast area


The myth of the 50/50 breast

Med Phys 2009, vol 36

MEAN BREAST DENSITY: 13.7 % -‐ 25.6 % 80% WOMEN: < 27% GLANDULAR 95% WOMEN: < 45 % GLANDULAR

«(based on the results obtained in our study), THE “50-‐50” BREAST IS NOT A REPRESENTATIVE MODEL OF THE BREAST COMPOSITION»


Volumetric methods

Ø  VBD is the ra7o between the fibroglandular volume and the breast volume

Ø  From a physical model of the mammography system (X-‐ray source, detector, scaPering, beam hardening), aPenua7on is calculated pixel-‐by-‐pixel

Ø  For each pixel, thicknesses of glandular 7ssue and fat are computed

Ø  For processing images required Ø  Volpara (Matakina), Quantra (Hologic), Cumulus V

(University of Toronto)


VBD: results

0 10 20 30 40 50 60 70 80 90 100 0 200 400 600 800

1000 1200 1400

VBD (%)

per-image

0 5 10 15 20 25 30 35 40 45 50 0

200

400

600

800

1000

1200

1400

VBD (%) ü  The scale is clearly not [0-‐100%] ü  Breast density is below 50% (min 1.47% -‐ max 47.2%) ü  Breast density is below 30% in 97.9% of cases ü  Mean:12.4% ü  Median: 10.5%

8433 immagini

PADUA, IOV •  Volpara 1.4.3 •  Images from GE DS, GE Essen@al, IMS Giouo Image 3DL


0 20 40 60 80 100 1200

5

10

15

20

25

30

35

40

45

50

VB

D (%

)

breast thickness (mm)

VBD: results PADUA, IOV •  Volpara 1.4.3 •  Images from GE DS, GE Essen@al, IMS Giouo Image 3DL

20 30 40 50 60 70 80 90 1000

5

10

15

20

25

30

35

40

45

50

mea

n V

BD

per

-pat

ient

(%)

Age (y)

Thicker breasts can be less dense Breast density tends to decreasewith pa7ent age


What is going on ?...

Enacted law

Introduced Bill

Working on Bill

No ac@on

Insurance coverage law

HEALTH JOURNAL August 6, 2012, 7:05 p.m. ET The Latest Mammogram Controversy: Density Many Women Aren't Told Their Breast Type May Cloud Cancer Screening; More States Consider No@fica@on

Wall Street Journal


hPp://areyoudenseadvocacy.org/dense/

1.  Breast density is associated to mammography sensi7vity (lesion detec7on) and to addi7onal cancer risk.

2.  Can be evaluated by human readers (subjec7vity) 3.  Can be quan7fied by digital mammography (percentage of

glandular components, either area or volume based). 4.  In the U.S. quan7fica7on of breast density is becoming

mandatory by law. 5.  APempt to define a threshold to have addi7onal imaging

(for example US) in screening.

Take home messages


Thank you !

[email protected]

advances)in)digital) mammography) - bem vindo ao sítio … 3/auditorium/2 advances in...

Documents