Low Dose Molecular Breast Imaging - AAPM: The American ... · Four dose reduction techniques have been developed and evaluated in order to reduce the administered dose of Tc-99m sestamibi

Michael O’Connor, Ph.DDept. of Radiology

Mayo Clinic

Low Dose Molecular Breast

Imaging

This work has been funded in part by the following:National Institute of Health

Dept. of DefenseSusan G Komen Foundation

Mayo FoundationFriends for an Earlier Breast Cancer Test

Conflict of InterestDr. M.K. O’Connor Royalties - Gamma Medica

Research funding – GE HealthcareResearch support – MTTI

• ACR BI-RADS (qualitative score)• Used by radiologists as part of routine practice• ACR BI-RADS (qualitative score)• Used by radiologists as part of routine practice

Breast Density ClassificationBreast Density Classification

<25%<25% 25-49%25-49% 50-74%50-74% ≥75%≥75%BI-RADS 1 BI-RADS 2 BI-RADS 3 BI-RADS 4

Sens = 88%Sens = 88% 82%82% 69%69% 62%62%

Breast DensityComparative Relative Risks

Breast DensityComparative Relative Risks

Risk factor Relative riskBRCA mutation 20Lobular carcinoma in situ 8-10Dense breast parenchyma 4-6Personal history of breast cancer 3-4Family history (1° relative) 2.1Postmenopausal obesity 1.5

Risk factor Relative riskBRCA mutation 20Lobular carcinoma in situ 8-10Dense breast parenchyma 4-6Personal history of breast cancer 3-4Family history (1° relative) 2.1Postmenopausal obesity 1.5

“Mammographic density is perhaps the most undervalued and underutilized risk in studies

investigating the causes of BC.”

Breast Density LegislationBreast Density Legislation

Connecticut enacted similar legislation in 2009 and a comparable breast density patient education bill, SB 173, is scheduled to be heard in the California

Assembly this month.

Sensitivity of Mammogram, US, and MRIin Women at Increased Risk

Sensitivity of Mammogram, US, and MRIin Women at Increased Risk

Subjects Sensitivity Sensitivity SensitivityAuthor/year Country (no.) MMG (%) US (%) MRI (%)

Kuhl, 2000 Germany 192 33 33 100

Warner, 2004 Canada 236 36 33 77

Kriege, 2004 Netherlands 1,909 40 NA 71

Kuhl, 2005 Germany 529 33 40 91

Leach 2005 U.K. 649 40 NA 77

Sardanelli, 2006 3571 40 43 81

Subjects Sensitivity Sensitivity SensitivityAuthor/year Country (no.) MMG (%) US (%) MRI (%)

Kuhl, 2000 Germany 192 33 33 100

Warner, 2004 Canada 236 36 33 77

Kriege, 2004 Netherlands 1,909 40 NA 71

Kuhl, 2005 Germany 529 33 40 91

Leach 2005 U.K. 649 40 NA 77

Sardanelli, 2006 3571 40 43 81

ACRIN 66663-Year Screening Study of Ultrasound and Mammography in 2500

High-Risk Women

41 cancers detected20 by mammography (sens 50%)20 by ultrasound (sens 50%)

Mammography: 3% biopsy, 29% positiveUltrasound: 5% biopsy, 9% positive

Study PI – Wendie Berg: "Women contemplating breast ultrasound screening mustbe aware of the substantial risk of false positives“

Study Co-I – Etta Pisano: “based on the study results, it is difficult to determine whether screening ultrasound is worthwhile”

Breast MRIPoor specificity (highly variable: 50% - 90%)Patient acceptance of MRI?

• ACRIN 6666 trial

• 42% declined free MRI (25% due to claustrophobia)

Expensive (~10 times that of mammography)

Molecular Imaging of the Breast

PEM (Positron Emission Mammography)• Coincidence detection using 2 scanning arrays of LYSO crystals

• Clinical unit developed by Naviscan

BSGI (Breast Specific Gamma Imaging)• Single detector, multicrystal NaI based gamma camera

• Developed by Dilon Technologies

MBI (Molecular Breast Imaging)• Dual detector Cadmium Zinc Telluride based gamma cameras

• Clinical units developed by Gamma Medica and GE Healthcare

PEM (Positron Emission Mammography)

FOV: 16 cm x 24 cm2 scanning arrays of LYSO

2.4 mm resolution (in plane)8.0 mm resolution (cross-plane)(JNM 2009;50:1666-1675)

Patient fasting 4-6 hr.Inject 10 mCi F-18 FDGWait ~60 minutesObtain CC / MLO viewsScan time ~10 minutes / view








BSGI (Breast Specific Gamma Imaging)

FOV: 15 cm x 20 cmSingle array of NaI crystals

3 mm intrinsic resolution~18% energy resolution

No patient fasting requiredInject 25-30 mCi Tc-99m sestamibiImage ~5 minutes post injectionObtain CC / MLO viewsScan time ~10 minutes / view








Breast Phantom: Comparison between Systems*Breast Phantom: Comparison between Systems*

Tumor Depth

1cm

3 cm

5 cm

MC-NaI NaI CZT MC-CsI

*Hruska CB, et al. Nucl Med Commun, 2005; 26: 441-445

Cadmium Zinc Telluride (CZT) Detector

• Excellent Intrinsic Resolution = 1.6 mm / 2.5 mm

• Excellent Energy Resolution 4.0% / 6.5%

•• Can be operated at room temp

• No dead space – ideal for breast imaging

• Expensive – currently limited to small field of view detectors

• First commercial systems using CZT developed for nuclear cardiology

2.54 cm

Molecular Breast Imaging Procedure

• Patient receives an IV injection of a radiotracer (Tc-99m sestamibi)

• The tracer preferentially accumulates in cancer cells and is not influenced by breast density

• The breast is lightly compressed between the 2 gamma cameras, only light pain-free compression is necessary

• Imaging starts ~5 minutes post injection. Acquire CC and MLO views of each breast for 10 minutes / view

MBI Systems Currently at Mayo Clinic

GE Research Unit GM-I Research Unit GM-I Clinical Unit

Can MBI find lesions not visible on mammography?

1.1cm nodular density in upper inner right breast 9.5cm from the nipple

Can MBI find lesions not visible on mammography?

2 x 1 cm IDC with multiple satellite lesions confirmed as DCIS at surgery

Comparison of Screening MBI and MammographyComparison of Screening MBI and Mammography

~1000 patients - asymptomaticCompare MBI and mammography in patients withdense breasts at increased risk of breast cancer

MBI performed with 20 mCi Tc-99m sestamibiBreast status assessed @ 12-months

Question – is MBI a viable screening adjunct to mammography in patients with dense breasts?

~1000 patients - asymptomaticCompare MBI and mammography in patients withdense breasts at increased risk of breast cancer

MBI performed with 20 mCi Tc-99m sestamibiBreast status assessed @ 12-months

Question – is MBI a viable screening adjunct to mammography in patients with dense breasts?

Diagnostic accuracy of screening mammography and MBI

75% (3/4)75% (3/4)50% (2/4)50% (2/4)25% (1/4)25% (1/4)Sensitivity (DCIS)Sensitivity (DCIS)

28% (10/36)28% (10/36)

7.8% (73/936)7.8% (73/936)

93% (861/925)93% (861/925)

100% (8/8)100% (8/8)

83% (10/12)83% (10/12)

MBI aloneMBI alone

24% (11/45)24% (11/45)

16% (146/936)16% (146/936)

85% (789/925)85% (789/925)

100% (8/8)100% (8/8)

92% (11/12)92% (11/12)

Mammography Mammography plus MBIplus MBI

25% (2/8)25% (2/8)Sensitivity Sensitivity (Invasive cancers)(Invasive cancers)

91% (839/925)91% (839/925)SpecificitySpecificity

9.5% (89/936)9.5% (89/936)Recall RateRecall Rate

18 % (3/17)18 % (3/17)PPVPPV

25% (3/12)25% (3/12)Sensitivity (all Sensitivity (all cancers)cancers)

Mammography Mammography alonealone

Rhodes et al, Radiology 2011;258:106-118

13.5 mm ILC (total extent 5.1 cm)

9 mm DCIS7 mm tubulolobular ca

17 mm IDC + DCIS

9 mm IDC9 mm ILC10 mm + 16 mm IDC

Mammographically occult cancers detected on MBI

Hendrick R.E.Radiation Doses and Cancer Risks from Breast Imaging StudiesRadiology. 2010 Oct; 257(1):246-53.

O'Connor MK, Li H, Rhodes DJ, Hruska CB, Clancy CB, Vetter RJ.Comparison of radiation exposure and associated radiation-induced cancer risks from mammography and molecular imaging of the breast.Med Phys. 2010 Dec;37(12):6187-98.

Relative Radiation RisksRadiation risk to patients

Mammogram ~ 0.5 mSvPEM (10 mCi F-18 FDG) ~ 7 mSvBSGI (25-30 mCi Tc-99m mibi) ~ 9 mSvMBI (20 mCi Tc-99m mibi) ~ 6 mSv

For pop. of 100,000 women undergoing above procedures at age 40, estimated cancer mortality

Mammogram ~ 2PEM (10 mCi F-18 FDG) ~ 30BSGI (25-30 mCi Tc-99m mibi) ~ 35MBI (20 mCi Tc-99m mibi) ~ 24

Where does the estimate of cancer

mortality come from ?

Based on Table 12D from BEIR VII

100,000 women aged 30

Single dose of

100 mGy

Over their lifetime

…range of plausible values for LAR is labeled a “subjective confidence interval” to emphasize its dependence on

opinions in addition to direct numerical observation (BEIR VII, page 278)

Risk Models

•• Lifetime Attributable Risk (LAR)Lifetime Attributable Risk (LAR)•• ““Because of the various sources Because of the various sources

of uncertainty it is important to of uncertainty it is important to regard specific estimates of LAR regard specific estimates of LAR with a healthy skepticism, placing with a healthy skepticism, placing more faith in a range of possible more faith in a range of possible valuesvalues”” (BEIR VII, page 278)(BEIR VII, page 278)

For pop. of 100,000 women exposed to naturally occurring background radiation from age 0-80, estimated cancer mortality

Background radiation

U.S. Average (3.1 mSv/year) ~1010

Colorado (~4.5 mSv/year) ~1460

Florida (~2.5 mSv/year) ~ 810

Relative Radiation Risks

0.00

2000.00

4000.00

6000.00

8000.00

10000.00

12000.00

14000.00

16000.00

18000.00

0 20 40 60 80 100

Age(years)

Cum

ulat

ive

Can

cer M

orta

lity

U.S. Cancer MortalityMinnesota - 3 mSv/yearColorado - 4.5 mSv/year50 mSv at age 40

Cancer mortality in 100,000 subjects

240 deaths

Estimated deaths due to background

radiation.No epidemiological evidence to support

these numbers

0

100

200

300

400

500

MBI Tech

General NMT

Cardiology NMT

Radiographer

mre

m

Ann

ual T

echn

olog

ist D

ose

MBI: Implications for Radiation Exposure to the Technologist from 8 patients / day, 20 mCi / patient

Pletta CK, et al. J Nucl Med 2009; 50 (Suppl 2): 428P.

Radiation Dose Reduction

Reduce dose / increase scan time ?- already at ~40 minutes !!

Improve the technology- detector / collimator optimization- energy window optimization- noise reduction algorithms- composite image from opposing detectors

Molecular Breast ImagingDetector / Collimator Optimization

Clinical studies show average breast thickness = 6 cmHence each collimator optimized over range 0 – 3 cm

Monte Carlo simulations (MCNP) used tooptimize detector and collimator design

What is the optimal pixel size and collimation to achieve a spatial resolution = 5 mm at 3 cm?

Optimal pixel size for near-field imaging (0-3 cm)

100%

65%

45%

0.30

Molecular Breast ImagingConventional Collimation

CZT: 1.6 x 1.6 mm pixel

Hexagonal holeCollimator:2.5 mm hole diameter

Molecular Breast ImagingConventional Collimation

CZT: 1.6 x 1.6 mm pixel

Square holeCollimator:~1.6 mm hole diameter

Molecular Breast ImagingCollimator specifications

Results from Simulation

Better resolution@ 3 cm

Factor of ~2 gain in sensitivity

expected

Molecular Breast ImagingMatched Collimation on CZT detectors

Tungsten collimator

Square holes

Each hole matched to individual pixel on

CZT detector

Molecular Breast ImagingEffect of Optimal Collimation – CZT Detector MBI – Radiation Dose Reduction

Four dose reduction techniques have been developed and evaluated in order to reduce the administered dose of Tc-

99m sestamibi needed for MBI

- collimator optimization

- energy window optimization- noise reduction algorithms

- composite image from opposing detectors

Wide energywindow

(110-154 kev)

~1.4 gain insensitivity

System sensitivity – effect of collimation and energy window

Detector Collimator Energy Window

counts/min/μCi

Relative gain in

counts/pixel

CZT1.6 mm

pixel

Standard Standard, 140 ± 10% 312 1.0

Standard Wide, 110 – 154 keV 391 1.3

Tungsten design Standard 140 ± 10% 905 2.9

Tungsten design Wide, 110 – 154 keV 1132 3.6

CZT2.5 mm

pixel

Standard Standard, 140 ± 10% 254 1.0

Standard Wide, 112 – 154 keV 331 1.3

optimized design Standard 140 ± 10% 534 2.1

optimized design Wide, 112 – 154 keV 705 2.8

GMI LumaGem@

1 cm

GMI LumaGem@

3 cm

Contrast to Noise Ratio – Lumagem 3200s Contrast to Noise Ratio – Discovery 750b

MBI Dose Reduction

Comparison of count densities in patient studiesSystem # Studies Ratio of Cts (New: Old)GM CZT 38 4.9 + 1.6

Dose

20 mCi

8 mCi

MBI Dose ReductionRelative Gain = 4.9 + 1.6

MBI – Radiation Dose Reduction

Four dose reduction techniques have been developed and evaluated in order to reduce the administered dose of Tc-

99m sestamibi needed for MBI

- collimator optimization

- energy window optimization

- noise reduction algorithms

- composite image from opposing detectors

Noise Reduction / Combined Images from Opposite Detectors

Combined Images:Gaussian Neighborhood Geometric Mean (GNGM)

Noise Reduction:Non-Local Means (NLM) denoising filterWide Beam Reconstruction (UltraSPECT)

Image Generation & analysisFilter parameters and order of application of noisereduction algorithms yet to be optimized

Gaussian Neighborhood / Geometric Mean Algorithm

8 mCiRaw data

4 mCiGNGM / NLM

4 mCiWBR

Comparison of Noise Reduction algorithms

Comparison of Low-dose MBI and Mammographyin a Screening Environment


Question – is low-dose MBI (~4 mCi) a viable screening adjunct to mammography in patients with dense breasts?

Compare MBI and mammography in asymptomatic patients with dense breasts on prior mammogram

2400 patients over next 2 years

1100 patients imaged as of 7/25/1114/15 cancers detected on MBI1/15 cancer detected on mammography1 cancer missed by MBI and mammography

Question – is low-dose MBI (~4 mCi) a viable screening adjunct to mammography in patients with dense breasts?

Compare MBI and mammography in asymptomatic patients with dense breasts on prior mammogram

2400 patients over next 2 years

1100 patients imaged as of 7/25/1114/15 cancers detected on MBI1/15 cancer detected on mammography1 cancer missed by MBI and mammography


Screening Mammogram(negative)


MBI (positive)



ScreeningMammogram

(negative)

MBI (20 mCi)(negative)

March 2008

MBI (4 mCi)(positive)

March 2011

2 cm IDC



ScreeningMammogram

(negative)


Ultrasound6.5 mm Invasive

Tubular carcinoma


ScreeningMammogram

(negative)


Ultrasound~3 cm Invasive

Lobular carcinoma

Comparison of Low-dose MBI and Mammography

Screening Mammogram(negative)

MBI (2 mCi)

MBIMultifocal DCIS with ADH

Advantages of MBI•• Compared to mammography:Compared to mammography:

•• Higher sensitivity in dense breastsHigher sensitivity in dense breasts•• Comparable radiation burden to the body Comparable radiation burden to the body •• Less compression, improved patient comfortLess compression, improved patient comfort

•• Compared to MRICompared to MRI•• Comparable sensitivity, better specificityComparable sensitivity, better specificity•• Offers option if contraindication to MRI Offers option if contraindication to MRI

(claustrophobia, pacemaker, clips)(claustrophobia, pacemaker, clips)•• Avoids gadolinium and potential for NSFAvoids gadolinium and potential for NSF•• Easy to interpret Easy to interpret –– 8 images vs 1000s for MRI8 images vs 1000s for MRI• 5-7 fold less expensive than MRI

Screening for Breast Cancer•• Mammography has proven to be an Mammography has proven to be an

excellent screening technique for the excellent screening technique for the majority of womenmajority of women

•• SubSub--optimal in women with dense breast optimal in women with dense breast tissuetissue

•• Alternative techniques are requiredAlternative techniques are required•• Possible techniques include ultrasound, Possible techniques include ultrasound,

MRI and lowMRI and low--dose molecular imagingdose molecular imaging•• Cost and specificity will be key factors to Cost and specificity will be key factors to

acceptance of any adjunct screening acceptance of any adjunct screening techniquetechnique

Documents

Low Dose Molecular Breast Imaging - AAPM: The American ... · Four dose reduction techniques have been developed and evaluated in order to reduce the administered dose of Tc-99m sestamibi