－W1－

Objectives and current status of QIBA (Quantitative Imaging Biomarkers Alliance)

Daniel C. SullivanDepartment of Radiology, Duke University Medical Center

Abstract　　A quantitative imaging biomarker (QIB) is an objectively measured characteristic derived from an in vivo image as an indicator of normal biological processes, pathogenic processes or response to a therapeutic intervention. In 2007 the Radiological Society of North America (RSNA) organized the Quantitative Imaging Biomarkers Alliance (QIBA) whose mission is to improve the value and practicality of quantitative imaging biomarkers by reducing variability across devices, patients and time.　　The QIBA initiative involves: (1) stakeholder collaboration to identify needs and solutions to develop consistent and reliable quantitative imaging results across imaging platforms, clinical sites, and time to achieve accurate and reproducible quantitative results from imaging methods. Since the process of acquiring a clinical imaging scan is complex, the goal requires much coordinated work among many stakeholders.　　There are several sources of variability in quantitative results from clinical images: (1) image acquisition hardware, software and procedures; (2) measurement methods; and (3) reader variability. QIBA employs a consensus-driven approach to produce a QIBA Profile that includes one of more QIBA Claims and specifications for the image acquisition necessary to achieve the QIBA Claim. QIBA Profiles are based on published data whenever such data are available and on expert consensus opinion where no data exist.　　Although based primarily in the USA, there are QIBA participants from North and South America, Europe and Asia. At the 2015 European Congress on Radiology, the European Society of Radiology (ESR) announced the formation of the European Imaging Biomarkers Alliance (EIBALL). In addition, leaders of the Japan Radiological Society (JRS) have met with the QIBA leaders to discuss future collaborations. Dr. Sullivan’s lecture at the Fall Meeting of the JRS in October 2015 will provide more details about QIBA activities.

Key wordsQuantitative Imaging Biomarkers Alliance (QIBA), Radiological Society of North America (RSNA), standardization, reproducibility, precision medicine

Rinsho Hyoka Clinical Evaluation

－W2－

1．Introduction

2．Problem of variation and wrong scan interpretation

－W3－

Fig. 1　Operating points of 108 radiologists reading same 100 mammograms

0.0 0.2 0.4 0.6 0.8 1.0FP

0.0

0.2

0.4

0.6

0.8

1.0

TP

Skill

ValueJudgments

－W4－

2.1　Premise of the problem

2.2　 Computers cannot replace radiologists

3．Motivation for QIBA

3.1　Quantitative imaging in healthcare

3.2　 Physician’s use of quantitative information

－W5－

Fig. 2　Motivation for QIBA

Treat

Wait

ClinicalDecision

QuantitativeBiomarker

Measure = 10 ±6

?!

8.5

3.3　Variability in imaging measurements

3.4　Variability in scanner measurements

－W6－

3.5　 Poor reproducibility ‒ Clinical implications

4．Quantification ‒ Consumer expectations

4.1　 Publications show need for more quantitative information

－W7－

4.2　 Research on quantitative imaging is increasing

－W8－

4.3　 Current commercially available MR QIB applications

5．Impediments in using quantitative imaging

5.1　Lung densitometry example

5.2　Toward quantitative imaging

Fig. 3　Toward quantitative imaging

Clinical Valueof QI Data

Data fromclinical trials

showingClinical Value

Accurate,reproducible QI

Data fromscanners

－W9－

6．Quantitative Imaging Biomarkers Alliance (QIBA)

6.1　QIBA organization chart

Fig. 4　QIBA organization chart

QIBA Steering CommitteeJackson (Chair)

Perlman (Vice-Chair)

CT Coordinating CmteGoldmacher, Schwartz, Lynch

CT VolumetryBiomarker CmteGoldmacher, Samei,

Siegelman

Volumetry AlgorithmChallenge TF

Athelogou

Small Lung Nodule TFGierada, Mulshine, Armato

QIBA/fNIH FDABiomarker

QualificationPartnership

Lung DensityBiomarker CmteLynch, Fain, Fuld

Airway MeasurementTF

Fain

NM Coordinating CmteWahl, Perlman, Mozley

FDG-PET BiomarkerCmte

Sunderland, Subramaniam,Wollenweber

Profile ConformanceTF

Turkington, Lodge,Boellaard

QIBA/fNIH FDABiomarker

QualificationPartnership

PET-AmyloidBiomarker Cmte

Smith, Minoshima, Perlman

SPECT Biomarker CmteSeibyl, Mozley, Dewaraja

Clinical LiteratureReview TF

Seibyl

ImageAcquisition &

Image Processingfor DaTscan TF

Dewaraja

Phantoms DROTF

Dickson, Zimmerman

Quantitative & ImageAnalysis TF

Miyaoka, Seibyl

MR Coordinating CmteGuimaraes, Zahlmann, Elsinger

PDF-MRI BiomarkerCmte

Rosen, Boss, Kirsch

DWI-MRI TFBoss, Chenevert

DCE-MRI TFLaue, Chung

DSC-MRI TFErickson, Wu

DTI TFProvenzale, Schneider

MRE Biomarker CmteEhman, Cole

MRE Profile-Writing TFEhman, Cole

Proton Density FatFraction Biomarker

CmteReeder, Sirlin

fMRI Biomarker CmtePetrella, DeYoe, Reuss

fMRI Bias TFVoyvodic

US Coordinating CmteHall, Garra

US SWS BiomarkerCmte

Hall, Garra, Milkowski

System Dependencies/Phantom Testing TF

Palmeri, Wear

Clinical ApplicationsTF

Samir, Cohen-Bacrie,Cosgrove

US Volume FlowBiomarker CmteFowlkes, Kripfgans(AIUM supported)

US CEUS BiomarkerCmte

Averkiou, Barr

Process CmteO’Donnell, Sullivan

QIDW Oversight CmteErickson

TF = Task Force

27-May-2016

Scientific Liaisons:CT: Andrew BucklerMR: Tom ChenevertNM: Paul KinahanUS: Paul Carson

External RelationsLiaison:Daniel Sullivan

－W10－

6.2　 QIBA meeting at RSNA annual meeting

6.3　QIBA approach

6.4　QIBA profiles

6.5　Bias and precision

－W11－

Fig. 5　FDG-PET SUV example

TrueValue

True andMeasured

Values

Bias

2-Sigma

MeasuredValues

Good Accuracy and Good Precision

Freq

uenc

y of

Mea

sure

men

ts

TrueValue

True andMeasured

Values

Bias

2-Sigma

MeasuredValues

Good Accuracy and Poor Precision

Freq

uenc

y of

Mea

sure

men

ts

TrueValue

True andMeasured

Values

Bias

2-Sigma

MeasuredValues

Poor Accuracy and Good Precision

Freq

uenc

y of

Mea

sure

men

ts

TrueValue

True andMeasured

Values

Bias

2-Sigma

2-Sigma 2-Sigma

2-Sigma 2-Sigma

MeasuredValues

Poor Accuracy and Poor Precision

Freq

uenc

y of

Mea

sure

men

ts

Fig. 6　Bias and precision

－W12－

6.6　QIBA profile structure

6.6.1　Examples of QIBA claim statements

－W13－

6.6.2　 Expected precision for alternate scenarios

6.6.3　 CT volumetry example ‒ coefficients of variation

－W14－

6.6.4　 QIBA profile: Actors, Activities, Requirements

6.6.5　Conformance to QIBA profile

7．MR variables

－W15－

8．QIBA funds

9．Dynamic contrast-enhanced MRI (DCE MRI)

Fig. 7　RSNA QIBA DCE-MRI Digital Reference Object (DRO)

(Barboriak)• Simulated T1 measurement data for range of S0 levels and added

noise levels• Simulated DCE measurement data for range of S0 levels• Simulated DCE measurement data for extended Tofts model

－W16－

10．Biomarker development

11．European Imaging Biomarkers Alliance (EIBALL)

12．QIBA-related activity in Japan

13．Importance of obtaining clinical precision data

－W17－

14．Conclusion

Standardization and software

Q

SullivanQ

Sullivan

－W18－

Commitment of industries

Q

Sullivan

－W19－

Importance of quantitative imaging for drug development

Q

Sullivan

In the era of big data and precision medicine

Q

－W20－

Sullivan

Reproducibility and standardization in early and late phases

Q

Sullivan

Visual radiologist

Q

Sullivan

－W21－

Q

Sullivan

Q

Sullivan

Relationship between UPICT and QIBA

Q

Sullivan

－W22－

Q