quantitative ct imaging of the lung grand hyatt san ... qct... · dose modulation in an...

75
Quantitative CT Imaging of the Lung Grand Hyatt San Antonio San Antonio, Texas March 15, 2014 Program Description This one-day course, co-sponsored by the Society of Thoracic Radiology and the Quantitative Imaging Biomarkers Alliance, will summarize the current state of knowledge regarding the role of quantitative CT of the lungs in diffuse lung disease. Learning Objectives At the completion of the course, the attendee will be able to: • Describe the technical parameters recommended for performing quantitative CT of the lung parenchyma in diffuse lung disease • Understand the role of CT as a biomarker of diffuse diseases, including COPD and lung fibrosis • Comprehend newer methods for quantification pulmonary vasculature and lung texture • Understand the emerging role of MRI in the lung Parenchyma Target Audience Thoracic radiologists, pulmonologists, imaging scientists. Credit Information The Society of Thoracic Radiology is accredited by the Accreditation Council for Continuing Medical Education to sponsor continuing medical education for physicians. The Society of Thoracic Radiology (STR) designates this live activity for a maximum of 6.0 AMA PRA Category 1 CreditsTM. Physicians should claim only credit commensurate with the extent of their participation in the activity. Course Directors David A. Lynch, MB and Jonathan Goldin, MBChB, PhD

Upload: vothu

Post on 17-Feb-2018

225 views

Category:

Documents


4 download

TRANSCRIPT

Page 1: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT Imaging of the Lung Grand Hyatt San Antonio

San Antonio, Texas March 15, 2014

Program Description This one-day course, co-sponsored by the Society of Thoracic Radiology and the Quantitative Imaging Biomarkers Alliance, will summarize the current state of knowledge regarding the role of quantitative CT of the lungs in diffuse lung disease.

Learning Objectives

At the completion of the course, the attendee will be able to: • Describe the technical parameters recommended for performing quantitative CT of the lung parenchyma in diffuse lung disease • Understand the role of CT as a biomarker of diffuse diseases, including COPD and lung fibrosis • Comprehend newer methods for quantification pulmonary vasculature and lung texture • Understand the emerging role of MRI in the lung Parenchyma

Target Audience

Thoracic radiologists, pulmonologists, imaging scientists.

Credit Information The Society of Thoracic Radiology is accredited by the Accreditation Council for Continuing Medical Education to sponsor continuing medical education for physicians. The Society of Thoracic Radiology (STR) designates this live activity for a maximum of 6.0 AMA PRA Category 1 CreditsTM. Physicians should claim only credit commensurate with the extent of their participation in the activity.

Course Directors

David A. Lynch, MB and Jonathan Goldin, MBChB, PhD

Page 2: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Program 7:30 AM - 8:00 AM Continental Breakfast Moderator: David A. Lynch, MB 8:00 AM - 8:10 AM Quantitative CT: Role of the Quantitative Imaging Biomarkers Alliance Daniel C. Sullivan, MD 8:10 AM - 8:30 AM What Do Clinicans Want from Imagers? George R. Washko, MD 8:30 AM - 8:50 AM QCT of the Airways Sean B. Fain, PhD 8:50 AM - 9:10 AM Clinical Evidence for CT as a Biomarker in COPD Jonathan Goldin, MBChB, PhD 9:10 AM - 9:30 AM QIBA Profile: Computed Tomography: Lung Densitometry Philip F. Judy, PhD 9:30 AM - 10:00 AM Discussion 10:00 AM - 10:20 AM Break Moderator: John D. Newell, Jr., MD 10:20 AM - 10:40 AM QCT in Lung Cancer Screening Pim de Jong, MD, PhD 10:40 AM - 11:00 AM Quantification of Pulmonary Vasculature on CT Raul San Jose Estepar, PhD 11:00 AM - 11:20 AM QCT in Lung Fibrosis David A. Lynch, MB 11:20 AM - 11:40 AM Texture Based Approaches to Image Analysis Joyce D. Schroeder, MD 11:40 AM - 11:50 AM Discussion 12:00 PM - 1:30 PM Sponsored Lunch: Workstation Demonstrations Moderator: Geoffrey D. Rubin, MD 1:30 PM - 2:30 PM Scientific Presentations Moderators: Jonathan Goldin, MBChB, PhD and Raul San Jose Estepar, PhD 2:30 PM - 2:50 PM MRI of the Lung Yoshiharu Ohno, MD, PhD 2:50 PM - 3:10 PM Break 3:10 PM - 3:30 PM Quantitative CT of the Lung in Normal Subjects Eric A. Hoffman, PhD 3:30 PM - 4:15 PM Panel Discussion: What is Necessary to Bring QCT into your Reading Room All Speakers

Carol
Typewritten Text
Carol
Typewritten Text
Room: Texas ABC
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 3: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Oral Presentations Moderators: Jonathan Goldin, MBChB, PhD and Raul San Jose Estepar, PhD

1 1:30 PM Size and Collapsibility of Emphysema Holes on CT of COPD: Evaluation with a New Method Joon Beom Seo, MD, PhD Seo JB, Oh SY, Lee M, Kim N, Lee SM

2 1:40 PM Automated Nodule Segmentation With Sub-voxel Accuracy Using Mutual Interaction of Pulmonary Segmentation Structures Ipek Oguz, PhD Oguz I, Raffy P, Wood S

3 1:50 PM Computer-aided Classification of Diffuse Lung Disease Opacities by Use of Sparse Representation-based Method on 3D-CT Images Shoji Kido, MD, PhD Kido S, Zhao W, Xu R, Hirano Y

4 2:00 PM Dose Modulation in an Anthropomorphic Chest Phantom and its Relative Effects when using Single and Dual Energy Scan Modes John D Newell Jr, MD Newell JD, Sieren JP, Guo J, Levy J, Hoffman EA

5 2:10 PM Lung Lesion Volume Measurements in Simulated Reduced-dose CT Scans Stefano Young, PhD Young S, Kim G, McNitt-Gray M

6 2:20 PM Determination of the Optimal Time-Window to Perform Wide Volume Dual Energy CT in a Swine Model of Pulmonary Embolism (PE) Laura Jimenez-Juan, MD Jimenez-Juan L, Dey C, Moghe S, Mehrez H, Homampour S, Paul NS

Scientific Poster Presentations

Bijan Image Quality Comparison of Low KV Imaging in CT Pulmonary Angiography Bijan B, Belashabadi S, Davoodi M, Kabir A

Jimenez-Juan

Dynamic Lung Perfusion with Wide Volume CT: Effect of Arterial Input on Arterial Flow Maps Jimenez-Juan L, Dey C, Homampour S, Merhez H, Paul NS

Jimenez-Juan

Dynamic Lung Perfusion with Wide Volume CT: Impact of Bolus Length on Arterial Flow Values Jimenez-Juan L, Dey C, Homampour S, Merhez H, Paul NS

Kicska Validation of Simulated Tomosynthesis: Can it Replace Real Imaging for First Line Clinical Trials? Kicska G

Scalzetti Feasibility of Cardiac Output Measurements Based on CT Timing Bolus Imaging Using a Modified Timing Bolus Scalzetti EM, Rajebi H, Ogden KM

Scalzetti Reproducibility of Cardiac Output Measurements Based on CT Timing Bolus Imaging Using a Modified Timing Bolus Scalzetti EM, Rajebi H, Ogden KM

Shim Distensibility Differs Regionally in Asthma vs. Normal Controls: A CT Investigation in SARP subjects Shim SS, Schiebler ML, Sorkness RL, Jarjour NN, Kanne J, Fain SB

Shim Airway Distensibility in Asthma Patients by Computed Tomography: Correlation with Clinical Indices and Reversibility after Bronchodilator Therapy Shim SS, Schiebler ML, Sorkness RL, Jarjour NN, Denlinger LC, Fain SB

Silva Lobar Contribution to Lung Volume and Heterogeneity in Healthy Subjects: Quantitative Assessment by CT Silva M, Nemec S, Dufresne V, Bankier AA

Page 4: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT:Role of the Quantitative Imaging

Biomarkers Alliance

D. Sullivan, MDDuke University;

RSNA

Why Must Imaging Become More Quantitative?

• Molecular medicine (personalized medicine)requires quantitative test results.

E id b d di i & QA P• Evidence-based medicine & QA Programsdepend on objective data.

• Decision-support tools (CADx, CDSS) need quantitative input.

Premise

• Variation in clinical practice results in poorer outcomes and higher costs.

RSNA’s Perspective:

• Extracting objective, quantitative results from imaging studies will g gimprove the value of imaging in

clinical practice.

ATS Policy Statement

An Official Research Policy Statement of the American Thoracic Society/European Respiratory Society: Standards for Quantitative Assessment of Lung Structure• Advances in CT technology have reduced the time for whole lung

imaging to 5 to 10 seconds, fueling a growing demand for rigorous validation of CT-derived quantitative measures in application to d /d i di ll f t d t tdrug/device discovery as well as safety and outcomes assessment.

• With the rapid progress in genome-wide searches, there is an additional need to use these quantitative measures along with characteristic pathology to establish disease phenotypes and to identify gene associations.

• 412 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 181 2010

Quantitative Imaging Biomarkers Alliance (QIBA): Background

• Started in 2007• Mission: Improve value and practicality of

quantitative imaging biomarkers by reducing variability across devices, patients, and time.y p

Carol
Typewritten Text
Quantitative CT: Role of the Quantitative Imaging Biomarkers Alliance Daniel C. Sullivan, MD Duke University
Page 5: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Collaborating Stakeholders

• Pharmaceutical companies• Imaging device companies• Imaging informatics companies• Government agencies• Professional societies• Clinical trialists and clinicians

QIBA Committees

Quantitative Magnetic Resonance Imaging [Q-MR] Perfusion, Diffusion, and Flow-MRI (PDF-MRI) Functional MRI (fMRI)

Quantitative Computed Tomography [Q-CT] CT Volumetry in Solid Tumors and Lung Nodules CT Densitometry in COPD Airway Morphology in Asthma

Quantitative Nuclear Medicine [Q-NM] FDG-PET SUV

Quantitative Ultrasound [Q-US] Shear Wave Speed for liver fibrosis

QIBA Approach

I. Identify Sources of Error and Variation.II. Specify Potential Solutions in the form of

QIBA “Profiles”QIBA Profiles .III. Test those Solutions. IV. Promulgate Solutions (Profiles) to Vendors

and Users.

1. Image acquisition variability2. Radiologist/Reader variability3 Measurement method variability

Variability in imaging measurements is related to:

3. Measurement method variability

10

Assays are characterized by their:• Technical Performance

Imaging Assays

• Clinical Performance Clinical validation Clinical utility

QIBA Profiles

A QIBA Profile describes a specific performance Claim and how it can be achieved.

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Quantitative CT: Role of the Quantitative Imaging Biomarkers Alliance Daniel C. Sullivan, MD Duke University
Page 6: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

QIBA Claim Template• List Biomarkers/Measurand(s)• Specify: Cross-sectional vs. Longitudinal

measurement• List Indices:

– Bias Profile (Disaggregate indices) – Precision Profile

• Test-retest Repeatability (Repeatability coefficient)• Reproducibility (Reproducibility coefficient; Intra-class

Correlation Coefficient [ICC]; Concordant Correlation Coefficient [CCC]).– Specify conditions, e.g.,

» Measuring System variability (hardware & software)

» Site variability» Operator variability (Intra- or Inter-reader)

• Clinical Context

True Biologic Change …

• … is approximately twice the test-retest variability

Cli i l Si ifi f th t h• Clinical Significance of that change needs to be determined by clinical studies.

Thank you.

Carol
Typewritten Text
Quantitative CT: Role of the Quantitative Imaging Biomarkers Alliance Daniel C. Sullivan, MD Duke University
Page 7: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

What Do Clinicians Want from Imagers?g

George Washko M.D.

Brigham and Women’s Hospital

2014

Consultancy Agreements

• Merck

• GlaxoSmithKline

• MedImmune

• Spiration

• Intellent Therapeutic Market Reports

A Clinician’s Basic Needs

• What does the patient have?

• How am I going to treat it?

Broad range of disability for any FEV1

Copyright ©2001 BMJ Publishing Group Ltd.

Jones, P W Thorax 2001;56:880-887

FEV1 35% Predicted FEV1 35% PredictedLongitudinal Change in FEV1

N Engl J Med 2011;365:1184-92

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
What do Clinicians Want from Imagers: George Washko, MD Brigham and Women's Hospital
Carol
Typewritten Text
Carol
Typewritten Text
Page 8: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

lobal Initiative for Chronicbstructivelobal Initiative for Chronicbstructive

GOGO

ungiseaseungisease

LDLD

© 2013 Global Initiative for Chronic Obstructive Lung Disease

Combined Assessment of COPD

k f Air

flow

Lim

itatio

n)

sk on h

isto

ry)

> 2 (C) (D)4

3

Ris

(GO

LD C

lass

ifica

tion

of Ris

(Exa

cerb

atio

1

0

(A) (B)

mMRC 0-1CAT < 10

2

1

mMRC > 2CAT > 10

Symptoms(mMRC or CAT score)

© 2013 Global Initiative for Chronic Obstructive Lung Disease

How can imaging help the clinician?

• Easy to use diagnostic

• Widely available

• Guides selection of therapy

• Immediately responsive to intervention

• Is cheaper and safer than trial/error

• Where to integrate imaging as a solution?

Emphysema and Lung Function

FEV1: 87% predicted

No spirometric obstruction

FEV1: 80% predicted

No spirometric obstruction

Why use imaging?

Carol
Typewritten Text
What do Clinicians Want from Imagers: George Washko, MD Brigham and Women's Hospital
Page 9: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Do we need imaging?

• Heterogeneous disease process

• Heterogeneous response to treatmentSpirometry– Spirometry

– Symptoms

– AECOPD

• Outcomes require large cohorts followed over long periods of time

Imaging needs to demonstrate utility

• Must overcome nihilistic attitude about cause of COPD 

• Provide new insight into disease pathogenesis and progressionp g

• Show clinical value– Does current battery of treatments warrant imaging?

– You will be doing the clinical trials• Engage Clinicians, NIH, industry

Imaging Solutions

• What is the mechanism for susceptibility?

– Innate lung structure

– Inflammation/repair

Wh t i bi k f i t t?• What are unique biomarkers of interest?

– Markers of pathobiology

– Druggable targets

5

6

7

8

9

10

Voxe

ls

Quantification of emphysema on chest CT

0

1

2

3

4

-1000 -900 -800 -700 -600 -500

HU

% V

Hayhurst Lancet 1984, Müller Chest 1988, Gould ARRD 1988, Gevenois AJRCCM 1995, Coxson AJRCCM 1999; Slide from H. Coxson

Disease Progression

• Progression of emphysema

– Accurate to 1 HU annually?

– Variability of inspiratory effort

Volume correction (conservation of lung mass)– Volume correction (conservation of lung mass)

• What about the changes in intraparenchymal blood volume due to respiratory?

Carol
Typewritten Text
What do Clinicians Want from Imagers: George Washko, MD Brigham and Women's Hospital
Page 10: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Therapeutic Implications: A1AT Deficiency

Comparison AAT-treated versus placebo-treated subjects using combined data from 2 randomized, double-blind, placebo-controlled trials (n=119)

Mean f/u 2.5 years

Annual decline -1.73 g/L/yr vs. -2.74 g/L/yr in AAT-treated vs. placebo-treated subjects

Stockley, et al. Respiratory Research 2010;11:136

Noble Gas MRI

Yablonski et al. Journal of Applied Physiology

Bronchiectasis in COPD – Selection Criteria

• Prevalence of bronchiectasis 57% (cohort n=201) )

• Bronchiectasis associated with increased risk of death HR 2.54

Martinez-Garcia, et al. AJRCCM 2013

Imaging Solutions

• You cannot develop a great tool and expect someone to use it– Standardize the product

– Engage clinicians and find out what they needg g y

– Carry out clinical trials with standard and image based endpoints

• Goal: Prove that it is better than the current standard

Carol
Typewritten Text
What do Clinicians Want from Imagers: George Washko, MD Brigham and Women's Hospital
Page 11: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

5

Why use imaging?

Biologic Therapy A

Biologic Therapy B

Can imaging prompt screening/treatment for another 

condition?

Atlas of COPD, Springer Science. 2008

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
What do Clinicians Want from Imagers: George Washko, MD Brigham and Women's Hospital
Carol
Typewritten Text
Page 12: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT of the Airways Sean B Fain, PhD University of Wisconsin - Madison

2012/316049. Epub 2012 Jan

Quantitative CT of the Airways

Sean B. Fain, PhD Departments of Radiology and Medical Physics

University of Wisconsin - Madison

Faculty Disclosures STR – 2014 / San Antonio

Sean B. Fain

• Personal financial interests in commercial entities that

are relevant to my presentation(s) or other faculty roles:

GE Healthcare Research Grant (Current)

Xemed, LLC Scientific Advisory Board/Consultant (Current)

Educational Goals

• Describe the technical parameters recommended for performing quantitative CT of the lung parenchyma in diffuse disease

• Understand the role of CT as a biomarker of diffuse diseases, including COPD and lung fibrosis

• Comprehend newer methods for quantification of pulmonary vasculature and lung texture

• Understand the emerging role of MRI in the lung parenchyma

What is the role for imaging? • There is a gap in our knowledge of disease

progression – Regional inflammatory biomarkers – Regional airway remodeling biomarkers

• Goal for non-invasive regional assessment – Phenotyping severity (cross-sectional) – Monitoring progression (longitudinal) – Quantifying spatial heterogeneity

Airways: The Extracellular Matrix (ECM)1

Courtesy Dr. Robert Brown, Johns Hopkins University

High Resolution X-ray CT: Bronchodilation

Normal Abnormal Normal

Mild/ Moderate

Severe

1A. Shifren, et al. J Allergy (Cairo). 2012;2012:316049. doi: 10.1155/ 19. FRC baseline FRC post-albuterol

Page 13: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT of the Airways Sean B Fain, PhD University of Wisconsin - Madison

6

*

Challenges: Sources of Variability Overview of Quantitative CT Task

• Acquisition Protocol

– Parameters and CT scanner type/vendor – Low radiation dose protocol desired

• Spatial resolution limitations

Multi-Detector Scanner

Slip Ring Enables

Fast Gantry Rotation

Multiple Detector Row Acquired Simultaneou

Up to 4cm (64 rows) coverage

X-ray source

– Overestimation of airway wall (esp. smaller airways of ~2 mm diameter)

– Limited representation of distal airway segments

Airway Segmentation and Measurement

Image Data ~0.5 GB 0.5 X 0.5 X 0.6-0.7 mm3

TLC and FRC Lung Volumes

lume Coverage Achieved with a helical

Acquisition

• Short Breath-Hold time (<10 s) • 3D Isotropic Spatial Resolution

Stack of CT-slices of the lungs

acquired in 5 s breath-hold

Typical Acquisition Parameters2

• Breath-hold Coaching – Desire to scan at a known lung volume—for example, TLC, FRC, or RV.

• Desire 3D sub-millimeter near-isotropic resolution in the x, y, and z axes. – Use an optimal reconstruction kernel – Patient positioned at isocenter – Short breath-hold

• 120 kV

• 0.984 pitch; Rotation Speed: 0.5 s

• 40 mm collimation

• Use the lowest possible x-ray dose that meets the needs of a given research study. – 50-200 mAs (~1.5-8.0 mSv)

Airway Measurement

Full Width and Half Maximum (FWHM)3-5 Limitations of FWHM

Tube Current Scaled to BMI

2Development of Quantitative Computed Tomography Lung Protocols. Newell, John; Jr MD, FACR; Sieren, Jered; BS, RTR; MR, CT; Hoffman, Eric. Journal of Thoracic Imaging. 28(5):266‐271, September 2013.

3Israel Amirav, Sandra S. Kramer, Michael M. Grunstein, and Eric A. Hoffman, J. Appl Physiol 1993 Nov;75(5):2239-50. 4Gregory G. King, Nestor L. Müller, Kenneth P. Whittail, Qing-Sang Xiang, and Peter D. Paré. Am J Respir Crit Care Med 161: 574-580 (2000). 5Nakano; Jonathan C. Wong; Pim A. de Jong; Lilliana Buzatu; Taishi Nagao; Harvey O. Coxson; W. Mark Elliott; James C. Hogg; Peter D. Paré; Am J Respir Crit Care Med 2005, 171, 142-146. 6Joseph M. Reinhardt, Neil D. D’Souza, and Eric A. Hoffman. IEEE Trans on Mediical Imagiing 16(6):820-827.

Model-Based Airway Measurement6 Wall Thickness Measures

in Severe Asthma7

Wall Area % Wall Thickness %

60 * 59 58

57

56 55

54

53 52

Normal Mild-Moderate Severe

19

18.5

18

17.5

17

16.5

16

15.5

15

Normal Mild-Moderate Severe

WA % vs. Normal: p = 0.003

vs. MM: p = 0.005

WT % vs. Normal: p = 0.031

vs. MM: p = 0.014

6Joseph M. Reinhardt, Neil D. D’Souza, and Eric A. Hoffman. IEEE Trans on Mediical Imagiing 16(6):820-827. 7Aysola, et al. “Airway remodeling measured by multidetector CT is increased in severe asthma and correlates with pathology.” Chest 134 (6 ) (2008): 1183-91.

Page 14: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT of the Airways Sean B Fain, PhD University of Wisconsin - Madison

WA

%

WA

%

Airway Segmentation8

Regional Airway Wall Morphology

Seeded Region Growing Algorithm

Inspiratory Image Segmentation without Intensity Correction

Segmentation with Intensity Correction

WA% difference coded by airway

Severe > Non-Severe

Segmentation

without Linearity

Constraint

Segmentation with Linearity

Constraint

Airway Skeleton 8Tschirren J, Hoffman EA, McLennan G, Sonka M. IEEE Trans Med Imaging. 2005 Dec; 24(12):1529-39.

9SB Fain et al., Am. J. Respir. Crit. Care Med. , 2009. 179: p. A5575.

Regional Airway Wall Morphology

Non-Severe > Normal

WA% difference coded by airway segment

9SB Fain et al., Am. J. Respir. Crit. Care Med. , 2009. 179: p. A5575.

Regional Airway Wall Morphology

Severe > Normal

WA% difference coded by airway segment

9SB Fain et al., Am. J. Respir. Crit. Care Med. , 2009. 179: p. A5575.

Baseline

Longitudinal Changes in Severe Asthma

Year 2

Ongoing Clinical Research Studies

0.70

0.68

0.66

0.64 *

* p = 0.003 Severe vs.. normal

Severe

Nonsevere

Normal

0.70

0.68

0.66

* 0.64

Severe

Nonsevere

Normal

• Quantitative measures of airways have

established application in asthma and COPD phenotyping

– Severe Asthma Research Program (SARP)7 0.62

0.60

0.58

p = 0.022 Severe vs. mild-to-moderate

† p = 0.039 Severe vs. normal p = 0.008 Severe vs. mild-to-moderate

3 4 5 6

Generation

0.62

0.60

0.58

* p = 0.006 Severe asthma vs.. normal

p = 0.003 Severe asthma vs.. mild-to-moderate

3 4 5 6

Generation

– Spiromics11

– COPD Gene12 7Aysola, et al. “Airway remodeling measured by multidetector CT is increased in severe asthma and correlates with pathology.” Chest 134 (6) (2008): 1183-91. 11Kim, Woo Jin, et al. "CT metrics of airway disease and emphysema in severe COPD." Chest 136.2 (2009): 396-404 12Washko, George R., et al. "Airway wall attenuation: a biomarker of airway disease in subjects with COPD." Journal of

10Witt C. et al., Am J Respir Crit Care Med 183;2011:A372. Applied Physiology 107.1 (2009): 185-191.

Page 15: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT of the Airways Sean B Fain, PhD University of Wisconsin - Madison

MTF

(v)

Software Packages - Translation13-15

13VIDA Diagnostics, Coralville, IA (http://www.vidadiagnostics.com/) 14Fraunhofer MeVIS, Bremen, Germany (http://www.mevis.fraunhofer.de) 15Airway Inspector, Brigham and Women's Hospital, Boston, MA (http://www.airwayinspector.org/)

QIBA COPD/Asthma Subcommittee Future Directions

• Improve consistency and accuracy – Protocol standardization – Improve spatial resolution protocols

• Reduce radiation dose – Longitudinal progression of disease with

multiple lung volumes acquired – Explore iterative statistical and model-based

reconstructions

Overestimation of Airway Wall COPDGene 2 Phantom

Histology

CT

2 mm diameter

16Yasutaka Nakano; Jonathan C. Wong; Pim A. de Jong; Lilliana Buzatu; Taishi Nagao; Harvey O. Coxson; W. Mark Elliott; James C. Hogg; Peter D. Paré; Am J Respir Crit Care Med 2005, 171, 142-146.

COPDgene Study Quality Assurance Phantom (CTP698) 1

Airway Diameter Thickness WA%

1 6 1.5 55.56 2 3 0.6 48.98 3 6 0.9 40.83 4 2.5 0.4 42.61 5 6 1.2 48.98 6 2.5 0.4 42.61

1The Phantom Laboratory, Salem, NY

Reconstruction kernel MTF Examples

Airway Measures with ASIR and Edge Kernel

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

Soft

Standard

Detail

Bone

Bone Plus

Edge

Lung

ASIR with Edge Kernel (100 mAs Half DFOV)

4 5

2.5 mm diameter “airway”

4

0.2

0.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0

v (cycles/cm)

Page 16: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT of the Airways Sean B Fain, PhD University of Wisconsin - Madison

Airway Profiles Adaptive Statistical Iterative Reconstruction (ASIR) vs. Filtered Back Projection (FBP)

Standard

17 Rodriguez, A. et al; RSNA Nov. 29 2012, Chicago, IL (rsna2012.rsna.org/search/event_display.cfm?em_id=12034515)

Summary Conclusions

• Airway measurement using MDCT has advanced to practice – Active in ongoing several large multi-center clinical research

studies Software for measurement commercial and academic – are available

• Challenges remain – Repeatability – Precision and accuracy of wall measures

• Especially for more distal airways (~2 mm in diameter)

– Improvement in the context of dose reduction • Iterative statistical and model based reconstructions show promise

Thank you.

Questions?

• Funding:

Acknowledgements

• Duke University

References [1] A. Shifren, et al. J Allergy (Cairo). 2012;2012:316049. doi: 10.1155/2012/316049. Epub 2012 Jan 19.

[2] Newell, John; Jr MD, FACR; Sieren, Jered; BS, RTR; MR, CT; Hoffman, Eric. Journal of Thoracic Imaging. 28(5):266‐271, September 2013.

– QIBA Contract: NIBIB-PB-EB-1010- 159-JKS

– The Hartwell Foundation – NIH/NHLBI R01 HL080412 – T32-CA009206 – Vertex Pharmaceuticals

Physics

– Kevin Johnson, PhD – Grzegorz Bauman, PhD – Robert Cadman, PhD – Dr. Jie Tang, PhD for IT and

Reconstruction Support

• Pulmonology – Ronald Sorkness, PhD – Nizar Jarjour, MD – Loren Denlinger, MD – Michael Rock, MD

– Bastiaan Driehuys, PhD – Scott Robertson – Sivaram Kaushik

• Radiology

– Scott Nagle, MD – Brad Maxfield, MD

• Nurse coordinator

– Jan Yakey, RN – Jennifer Swartz, RN

• Research Technologists

– Kelli Hellenbrand, RT – Sara Pladziewicz, RT – Alice Minx, RT

[3] ISRAEL AMIRAV, SANDRA S. KRAMER, MICHAEL M. GRUNSTEIN, AND ERIC A. HOFFMAN, J. Appl Physiol 1993 Nov;75(5):2239-50.

[4] GREGORY G. KING, NESTOR L. MÜLLER, KENNETH P. WHITTALL, QING-SAN XIANG, and PETER D. PARÉ. Am J Respir Crit Care Med 161: 574-580 (2000).

[5] Nakano; Jonathan C. Wong; Pim A. de Jong; Lilliana Buzatu; Taishi Nagao; Harvey O. Coxson; W. Mark Elliott; James C. Hogg; Peter D. Paré; Am J Respir Crit Care Med 2005, 171, 142-146.

[6] Joseph M. Reinhardt,* Member, IEEE, Neil D. D’Souza, and Eric A. Hoffman. IEEE Trans on Mediical Imagiing 16(6):820-827.

[7] Aysola, et al. “Airway remodeling measured by multidetector CT is increased in severe asthma and correlates with pathology.” Chest 134 (6) (2008): 1183-91

[8] Tschirren J, Hoffman EA, McLennan G, Sonka M. IEEE Trans Med Imaging. 2005 Dec;24(12):1529-39. [9] SB Fain et al., Am. J. Respir. Crit. Care Med. , 2009. 179: p. A5575. [10] Witt C. et al., Am J Respir. Crit. Care Med 183;2011:A372. [11] Kim, Woo Jin, et al. "CT metrics of airway disease and emphysema in severe COPD." Chest 136.2 (2009): 396-404 [12] Washko, George R., et al. "Airway wall attenuation: a biomarker of airway disease in subjects with COPD." Journal

of Applied Physiology 107.1 (2009): 185-191. [13] VIDA Diagnostics, Coralville, IA (http://www.vidadiagnostics.com/) [14] Fraunhofer MeVIS, Bremen, Germany (http://www.mevis.fraunhofer.de) [15] Airway Inspector, Brigham and Women's Hospital, Boston, MA (http://www.airwayinspector.org/) [16] Yasutaka Nakano; Jonathan C. Wong; Pim A. de Jong; Lilliana Buzatu; Taishi Nagao; Harvey O. Coxson; W. Mark

Elliott; James C. Hogg; Peter D. Paré; Am J Respir Crit Care Med 2005, 171, 142-146. [17] Rodriguez, A. et al; RSNA Nov. 29 2012, Chicago, IL

(rsna2012.rsna.org/search/event_display.cfm?em_id=12034515)

Page 17: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Clinical Evidence for CT as a Biomarker Jonathan Goldin, MD, PhD DGSOM at UCLA
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 18: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Clinical Evidence for CT as a Biomarker Jonathan Goldin, MD, PhD DGSOM at UCLA
Page 19: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Clinical Evidence for CT as a Biomarker Jonathan Goldin, MD, PhD DGSOM at UCLA
Page 20: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Clinical Evidence for CT as a Biomarker Jonathan Goldin, MD, PhD DGSOM at UCLA
Page 21: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

��

Carol
Typewritten Text
Clinical Evidence for CT as a Biomarker Jonathan Goldin, MD, PhD DGSOM at UCLA
Carol
Typewritten Text
Page 22: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

QIBA Profile: Computed Tomography: Lung Densitometry 

Philip F. Judy, PhD

Brigham and Women’s Hospital

andand

Harvard Medical School

Boston, [email protected]

ChairRSNA QIBA COPD/Asthma Technical 

Committee

COPD/Asthma Technical CommitteeLung parenchyma density profile  

• A QIBA Profile is a document that describes a specific performance claim and how it can be achieved.Details : http://qibawiki.rsna.org/index.php?title=Main_Page

• Primary claim of the lung parenchyma density profile will be a consistency performance claim (1 HU, which i li 10% i t i “ l” f ti l thimplies 10% consistency in “normal” fraction less than ‐950 HU)

• Investigators have reported inconsistent lung parenchyma density measurements associated with:– scanner model– reconstruction protocol– radiation dose (mAs)– slice thickness.

Selected Inconsistency Reports• Scanner Model:  B.C. Stoel, F. Bode, A. Rames, S. Soliman, J.H. Reiber, J. Stolk, 

"Quality control in longitudinal studies with computed tomographic densitometry of the lungs," Proc Am Thorac Soc 5, 929‐933 (2008).

• Reconstruction protocol: K.L. Boedeker, M.F. McNitt‐Gray, S.R. Rogers, D.A. Truong, M.S. Brown, D.W. Gjertson, J.G. Goldin, "Emphysema: effect of reconstruction algorithm on CT imaging measures," Radiology 232, 295‐301 (2004). 

• Radiation Dose (mAs):  R. Yuan, J.R. Mayo, J.C. Hogg, P.D. Pare, A.M. McWilliams, S. Lam, H.O. Coxson, "The effects of radiation dose and CT manufacturer on measurements of lung densitometry," Chest 132, 617‐623 (2007).

• Slice Thickness:  D.S. Gierada, A.J. Bierhals, C.K. Choong, S.T. Bartel, J.H. Ritter, N.A. Das, C. Hong, T.K. Pilgram, K.T. Bae, B.R. Whiting, J.C. Woods, J.C. Hogg, B.A. Lutey, R.J. Battafarano, J.D. Cooper, B.F. Meyers, G.A. Patterson, "Effects of CT Section Thickness and Reconstruction Kernel on Emphysema Quantification Relationship to the Magnitude of the CT Emphysema Index," Acad Radiol 17, 146‐156 (2010)

QIBA Profile Elements for Discussion

1. Specialized QC Phantom – Water phantom is not sufficient1. J.P. Sieren, J.D. Newell, P.F. Judy, D.A. Lynch, K.S. Chan, J. Guo, E.A. Hoffman, 

"Reference standard and statistical model for inter‐site and temporal comparisons of CT attenuation in a multicenter quantitative lung study," Med Phys 39, 5757‐5767 (2012).

2. Breathing coaching and instructions3 Pixel noise specified3. Pixel noise specified

a. While pixel noise biases measurement ‐ constant tube current is specified  (Pixel standard deviation = 20 HU, approximately 50 mAs)

b. No AEC c. No dose reduction reconstructionsd. Rather, pixel noise is precisely specified 

HPR01 - LightSpeed 16

-1020-1015-1010-1005-1000-995-990-985-980

3/28/2009 1/22/2010 11/18/2010

Date

CT

Num

ber Inside Air

Outside Air

Lung - HPR01 - LightSpeed 16

-875

-870

-865

-860

-855

-850

-845

-840

-835

3/28/2009 10/14/2009 5/2/2010 11/18/2010

Date

Lung

CT

Num

ber

1.  Specialized QC Phantom Water CT number unlikely to identify important changes associated with recalibration 

Date

Material ShiftInside Air -19.78Outside Air -19.94Lung -18.05Water -0.96Acrylic -0.30

Inside air after = -1013.4

Note: This recalibration was an “over-correction”.

Carol
Typewritten Text
Carol
Typewritten Text
QIBA Profile: Computed Tomography: Lung Densitometry Philip F. Judy, PhD Brigham and Women's Hospital Harvard Medical School
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 23: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Air CT Numbers Inside and Outside COPDGene PhantomThree Clusters

GE LightSpeed16, Brilliance 64, Other Scanners

-1000

-995

-990

T N

umbe

r

Brilliance 64Definition

f S

-1015

-1010

-1005

-1010 -1005 -1000 -995 -990 -985 -980

Air Outside CT Number

Air

Insi

de C

T Definition AS+LightSpeed VCTSensation 16Sensation 64GE 16 Slice

-1000

-998

-996

-994

CT

Num

ber

Brilliance 64DefinitionDefinition AS+

Each point is data from a single scanner in the COPDGene StudyConcluded: 1. For these scanner  models using outside air CT number will not be useful correction for these CT scanner variations.  2. Intra‐model performance is small.

-1008

-1006

-1004

-1002

-1008 -1006 -1004 -1002 -1000 -998 -996 -994

Air Outside CT Number

Air

Insi

de C

LightSpeed VCTSensation 16Sensation 64

2.  Breathing coaching and instructions• For this scan, I am going to ask you to take a couple of 

deep breaths in and out before we have you breathe all the way in and hold your breath. 

• OK, let’s get started, • Take a deep breath in (watch chest to ensure a deep breath 

in) • Let it out (watch chest to ensure air is out) • Take a deep breath in (watch chest to ensure a deep breath 

in) • Let it out (watch chest to ensure air is out) • Now breathe all the way IN...IN...IN (watch chest to ensure 

a deep breath in as far as possible) • Keep holding your breath – DO NOT BREATHE! • Count 10 seconds (if practice) or scan; • At end of scan or practice: Breathe and relax

Pixel standard deviation of water and acylicregions of the COPDGene Phantom

• Data from Sieren JP, Newell JD, Judy PF, Lynch DA, Chan KS, Guo J, Hoffman EA., Reference standard and statistical model for inter site and temporal

3.  Pixel noise specified : Background

statistical model for inter‐site and temporal comparisons of CT attenuation in a multicenter quantitative lung study. Med Phys. 2012 Sep;39(9):5757‐67.

• Expiration protocol – 50 mAs , 120 kVp

• 47 scanner models

• 331 scans

y = ‐9.02x + 26.8R² = 0.27

20

25

30

Deviation [HU]

Water

COPDGene Phantom  ‐ 50mAs47 CT Scanners

3.  Pixel noise specified : Background

y = ‐11.62x + 28.4R² = 0.39

0

5

10

15

0.4 0.5 0.6 0.7 0.8 0.9 1

Pixel Standard D

Slice Thickness [mm]

Water

Acylic

Linear (Water)

Linear (Acylic)

y = 1.01x ‐ 0.56R² = 0.89

22

24

26

eviation [HU]

COPDGene Phantom  ‐ 50mAs47 CT Scanners

3.  Pixel noise specified : Background

14

16

18

20

14 16 18 20 22 24 26

Acrylic Pixel Standard De

Water Pixel Standard Deviation [HU]

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
QIBA Profile: Computed Tomography: Lung Densitometry Philip F. Judy, PhD Brigham and Women's Hospital Harvard Medical School
Carol
Typewritten Text
Page 24: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

22

24

26

Deviation [HU]

Brilliance 64

Definition

COPDGene Phantom  ‐ 50mAsSelected  CT Scanners

At least 3 scanners of same model

3.  Pixel noise specified : Background

14

16

18

20

14 16 18 20 22 24Water Pixel Standard 

Acrylic Pixel Standard Deviation [HU]

LightSpeed VCT

LightSpeed16

Sensation16

11

12

13

Pixel Standard 

n [HU]

Brilliance 64

COPDGene Phantom  ‐ AcrylicSelected  CT Scanners

At least 3 scanners of same model

3.  Pixel noise specified : Background

8

9

10

14 16 18 20 22 24 26

200 m

As Acrylic P

Deviation

50 mAs Acrylic Pixel Standard Deviation [HU]

Definition

LightSpeed VCT

LightSpeed16

Sensation 16

An approach to evaluate the effect of the variation of pixel standard deviation on fraction less than ‐950 HU

• Model of histogram – two components– Lung has two classes of pixels: Parenchyma and NonParenchyma

P h i l h G i hi t– Parenchyma pixels have a Gaussian histogram

– All NonParenchyma pixels are greater than the mode of lung histogram

• Sigma of Gaussian of parenchyma pixel has two components– Biological and measurement noise

– Measurement noise is pixel standard deviation

200000

250000

300000m Value

Case measured with Brilliance 64 – Fraction less than ‐950 = 7.8%Gaussian sigma = 41 HURed curve is lung parenchymaBlue curve is lung non‐parenchyma (partial volume of vessels

0

50000

100000

150000

‐1050 ‐950 ‐850 ‐750 ‐650 ‐550

Histogram

CT Number [HU]

Histogram

Fit Gaussian

Difference Above Mode

60

70

80

90

Fit [HU]

3.  Pixel noise specified : Background31 Cases – Same CT scannerIncrease BMI reduces detected x‐rays leading to higher noise

y = 1.3832x + 5.5912R² = 0.3402

0

10

20

30

40

50

0 10 20 30 40 50

Sigm

a of Gassian

 F

Body Mass Index

Sigma

Linear (Sigma)

150000

200000

250000

300000

gram

 Value

Sigma=41

Sigma=51

Sigma = 41  Fraction less than ‐950 = 7.8%Scanner with a pixel standard deviation 10 HU larger=> Sigma = 51  Fraction less than ‐950 = 11.0%

0

50000

100000

150000

‐1050 ‐950 ‐850 ‐750 ‐650

Lung Histog

CT Number [HU]

g

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
QIBA Profile: Computed Tomography: Lung Densitometry Philip F. Judy, PhD Brigham and Women's Hospital Harvard Medical School
Carol
Typewritten Text
Carol
Typewritten Text
Page 25: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

An Acquisition Specification Table from the Draft Profile

2.2 Radiation Exposure Specifications  CT acquisition shall use a tube current –rotation time product such that the pixel standard deviation measured in the lung test object described in IV. Compliance is less than 21 HU and greater than 19 HU. 

This is a nominal 50 mAs scan. COPDGene Phantom Data suggests that the specified mAs will range from 72 mAs to 32 mAs.

1. Subject Handling1.1  Contrast Preparation and Administration, no contrast agent1.2  Subject Positioning, e.g., table height  (3 DICOM Fields)1.3  Breathing Instructions during Acquisition

2. Image Data Acquisition2.1  Spatial‐Temporal Scanning Specifications, e.g., single slice thickness and pitch  

(7 DICOM Fields)

Tables in Image Acquisition Section from the Draft Profile 

2.2  Radiation Exposure Specifications  (6 DICOM Fields)2.3  Informational Specifications, e.g., vendor model and serial number (4 DICOM 

Fields)

3. Image Data Reconstruction , e.g. slice thickness and reconstruction algorithm (6 DICOM Fields)

QIBA Profile Elements for Discussion

1. Specialized QC Phantom – Water phantom is not sufficient

2. Breathing coaching and instructions

3 i l i i l ifi d3. Pixel noise precisely specified

4. Attempted to harmonize acquisition specification with CT air way size and CT lung nodule requirements.

5. Audience issues.

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
QIBA Profile: Computed Tomography: Lung Densitometry Philip F. Judy, PhD Brigham and Women's Hospital Harvard Medical School
Carol
Typewritten Text
Carol
Typewritten Text
Page 26: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

CT-based screening of the chest

Q tifi ti b d l- Quantification beyond lung cancer -

Pim A de Jong, Utrecht, the Netherlands

[email protected]

Take home points

1. Arterial, pulmonary and skeletal diseases can be quantified on

lung cancer screening chest CT

2. Cardiovascular events, lung cancer and osteoporotic events

can be predicted by using these quantitative measures

3. RCT multiple disease screening?

• screening of humans and not lung cancer

Lung cancer Emphysema

Bronchitis

Bronchiolitis

ILD

Problems in humans who smoke …

Systemic

inflammation

Cardio-vascular disease

OsteoporosisMetabolic syndrome

CT X-thoraxCardiovascular 3.37 3.28Lung cancer 2.96 3.51

Mortality in NLST

gOther cancer 2.89 3.08Respiratory 1.21 1.58

NEJM, 2011 Aug 4;365(5):395-409.

per 1000 personyears

Vertebral fracture assessment

Comprehensive CT quantification

osteoporosis

Lung segmentation and densitometry

Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Carol
Typewritten Text
Carol
Typewritten Text
Page 27: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

CT emphysema

IN 1 35%IN-950 = 1.35%

CT Air trapping

E/I ti 0 82E/I-ratioMLD = 0.82

Airway analysis

Fraunhofer MEVIS and DIAG Nijmegen

Cross-section every 1 mm along centerline

On average 1280 measurements per CT

Fraunhofer MEVIS and DIAG Nijmegen

CT bronchial wall thickness

Pi10 = 2.41 mm

Coronary and aortic calcifications

Isgum I, Med Phys 2010; 37(2):714-723

Carol
Typewritten Text
Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Carol
Typewritten Text
Carol
Typewritten Text
Page 28: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Vertebral fractures and bone density

Buckens CF, JBMR 2013 Jul 2

Why comprehensive CT quantification ?

1. reduce effort and variability

2. multiple disease screening

3. adjust screening intensity to individual risk

PROGNOSTIC VALUE OF ARTERIAL CALCIFICATIONS ON LUNG CANCER

SCREENING CT

Mets OM JACC Cardiovasc Imaging. 2013 Aug;6(8):899-907.

Derivation of a prediction model

1. Single center from the NELSON trial

2. N=1834 males, 60 years, 38 packyears

3. Age, smoking status, packyears, coronary calcium volume, aortic

calcium volume, cardiovascular history

4 From the baseline CT participants were followed for 3 years for CVD4. From the baseline CT participants were followed for 3 years for CVD

events (145 events)

5. Cox regression

6. Bootstrapping + shrinking

7. Internally validated prediction model

Validation of the prediction model

1. Another center from the NELSON trial

2. N=1725 males, 60 years, 38 packyears

3. Followed for 3 years: 118 CVD events

4. Calibration was good

5 Discrimination C index 0 71 (95%CI 0 67 0 76)5. Discrimination C-index 0.71 (95%CI, 0.67 – 0.76)

Hazard Ratios

Coronarycalcium

HR (95%CI)* Aortic calcium

HR (95%CI)*

100 mm3 1.08 (1.05 – 1.11) 100 mm3 1.02 (1.00 – 1.03)

250 mm3 1.21 (1.13 – 1.31) 250 mm3 1.04 (1.01 – 1.07)

500 mm3 1.48 (1.27 – 1.72) 500 mm3 1.08 (1.02 – 1.15)

1000 mm3 2.18 (1.61 – 2.94) 1000 mm3 1.17 (1.04 – 1.33)

≥ 1500 mm3 3.22 (2.05 – 5.05) 1500 mm3 1.27 (1.06 – 1.53)

2500 mm3 1.50 (1.10 – 2.04)

≥ 4000 mm3 1.90 (1.16 – 3.12)

* In multivariate analysis with age, smoking status, smoking history and cardiovascular history

Carol
Typewritten Text
Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Carol
Typewritten Text
Page 29: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

3-years risk calculator CVD risk according to the model

non High-risk (<6%)

High-risk (≥6%)

3-year risk [%], median (P25-P75) 3.4 (2.4 – 4.5) 10.3 (7.7 – 16.3)

Cardiovascular events N (%) 46 (4 0) 72 (12 2)Cardiovascular events, N (%) 46 (4.0) 72 (12.2)

No medication, % 68.3 42.1

Antihypertensive drugs only, % 12.9 15.7

Statins only, % 6.5 5.0

Both, % 12.2 37.2

Osteoporosis measures on lung cancer screening CT scans and all-cancer screening CT scans and all-

cause mortality

Stan Buckens, JBMR 2013 Jul 2

Are vertebral fractures or bone density independent predictors of mortality ?

Case-cohort osteoporosis study in NELSON

• Dual center longitudinal case-cohort study

• 3673 lung cancer screening trial participants

• Cases 196 deaths median follow-up of 6 years

• Random sample of N=394 from the cohort

• Presence and severity of vertebral fractures

• Bone density (Hounsfield Units) in Thoracic 12 or Lumbar 1

• Age, smoking details, coronary and aorta calcification volume

and pulmonary emphysema

• Cox proportional hazards modeling

Osteoporosis and all-cause mortality

Crude Fully adjusted*

fracture yes/no 1.93 (1.37 - 2.73) 2.18 (1.47 - 3.22)

worst fracture grade 1 1.93 (1.33 - 2.80) 2.12 (1.38 - 3.25)worst fracture grade 1 1.93 (1.33 2.80) 2.12 (1.38 3.25)

worst fracture grade 2 -3 1.97 (1.11 - 3.48) 2.37 (1.28 - 4.39)

cumulative fracture grade 1-3 1.89 (1.31 - 2.72) 2.13 (1.40 - 3.24)

cumulative fracture grade ≥4 2.16 (1.38 - 4.09) 2.37 (1.18 - 4.77)

vertebral attenuation 1.087 (1.027 – 1.15) 1.076 (1.006 - 1.15)

Carol
Typewritten Text
Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Carol
Typewritten Text
Carol
Typewritten Text
Page 30: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Risk for a future hip fracture

Buckens CF, JBMR 2013

Conclusion osteoporotic measures

• Vertebral fractures and bone density are associated with all-

cause mortality

• independent of age, smoking characteristics, coronary and

aortic calcifications and emphysema in male lung cancer

screening participants

Diagnosis of COPD in lung cancer screening CT

Objective

Can low-dose lung cancer screening CT be used to identify participants with h i b t ti l di ?chronic obstructive pulmonary disease ?

Mets et al. (2011) Identification of COPD in lung cancer screening CT scans. JAMA 306(16):1775-1781

Methods

N=1173 Nelson Utrecht, CT/PFT same day

Current or former heavy smokers 50 -75 years; smoking history >16.5 pysy ; g y py

COPD prebronchodil FEV1/FVC < 0.70723 normal lung function283 mild obstruction (FEV1 >80%)167 FEV1 <80%

Study populationMale, n (%) 1140 (97.2)

Age, years 62.1 ± 5.2

BMI, kg/m2 27.0 ± 3.6

Packyears 40 9 ± 17 9Packyears 40.9 ± 17.9

Current smoker, n (%) 629 (53.6)

Former smokers, n (%) 544 (46.4)

FEV1, %predicted 94.7 ± 17.6

FEV1/FVC, % 70.9 ± 9.4

Airflow obstruction, n (%) 450 (38.4)

Carol
Typewritten Text
Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Page 31: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Performance of chest CT

ACC(%)

SENS(%)

SPEC(%)

PPV(%)

NPV(%)

Baseline model 66.2 28.6 89.6 63.1 66.9

+ CT emphysema78 2 62 7 87 9 76 3 79 1

+ CT air trapping78.2 62.7 87.9 76.3 79.1

+ CT emphysema + CT-AWT

80.6 70.7 86.8 76.9 82.7

+ CT emphysema + CT-AWT + CT air trapping

82.8 73.2 88.8 80.2 84.2

Mets OM et al. Respiratory Research. 2013 May 27;14:59

Performance of CT in symptomatics

C-index (95% CI)

Asymptomatics Symptomatics

Baseline model 0 674 (0 625 0 722) 0 634 (0 589 -Baseline model 0.674 (0.625 -0.722) 0.634 (0.589 -

0.679)

+ CT emp BWT 0.828 (0.790 – 0.866) 0.886 (0.859 -

0.912)

+CT emp BWT Air T 0.832 (0.795 – 0.869) 0.905 (0.881 -

0.929)

Comprehensive CT quantification in lung cancer screening

1. Aiding detection of pulmonary nodules (first reader?)2. Quantification volume/mass pulmonary nodules3. Prediction of cardiovascular events4. Prediction osteoporotic fractures5. Diagnosis of COPD6. Prediction of lung cancer risk

Carol
Typewritten Text
Carol
Typewritten Text
QCT in Lung Cancer Screening Pim A. de Jong, MD, PhD University Medical Center Utrecht
Carol
Typewritten Text
Page 32: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantification of Pulmonary Vasculature on CT

Raúl San José Estépar1, Ph.D.

Department of Radiology, Brigham and Women’s Hospital, Boston MA  

Disclosures

• No disclosures

BackgroundPulmonary Vascular Disease

• Estimated prevalence of disease 26–70%1,2

• Presence associated with – Increased health care utilization3

– Increased mortality4

• Pathologic changes seen even in smokers with normal lung functionfunction

1. NEJM 1972;286:912-8

2. AJRCCM 2002;166:314-22

3. AJRCCM 1999;159:158-64

4. Am Rev Resp Dis 1979;119:895-902

Three Key Players in Vascular Related Smoking Pathology

Vessel RemodelingVessel Remodeling

InflammationInflammation

Changes are seen in Changes are seen in ““healthyhealthy”” heavy heavy 

smokers NOT just those with smokers NOT just those with 

quantifiable airways obstructionquantifiable airways obstruction

Endothelial Endothelial dysfunctiondysfunction

Imaging Landscape for Assessment of Pulmonary Vascular Disease

Static Vessel 

Morphology

Perfusion

Adaptive VQ 

Matching

Other Imaging 

Phenotypes

Genetics, Biomarkers

Mathematical Models: Enhance Value of Imaging 

Biomarker Personalized Disease Model Incorporating Pulmonary 

Vascular Network Dynamics

Why pulmonary vascular morphology?

• Static assessment of V/Q

• Potentially more sensitive marker of disease than airways or parenchyma biomarkers

• Evolution of central vasculature morphology• Evolution of central vasculature morphology that has shown to be highly statistically significant predictor of AECOPD1

• Largely unexplored

1 Wells et al. NEJM, 2012

Carol
Typewritten Text
Carol
Typewritten Text
Quantification of Pulmonary Vasculature on CT Raúl San José Estépar, Ph.D. Brigham and Women's Hospital
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 33: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Slide provided by Philippe Grenier

Prior Work

• Vascular pruning has been observed during angiography and histologic analysis in patients with PAH 

• Less than a handful of studies using imaging to tif th d f h i h lquantify the degree of changes in morphology 

of pulmonary vasculature 

Matsuoka et al, Am J Respir Crit Care Med. 2010;181(3):218-225

Fractal Dimension 

Moledina et al, Heart 2011 Aug;97(15):1245-9

CT Assessment of Vascular MorphologyCT Scan Lung and lobe extraction

Vessel Enhancing FilterScale Space Particles

Methods

• BV(CSA): Blood volume distribution as a function of CSA

• p(CSA): CSA probability• p(CSA): CSA probability density function – Kernel density estimation from 

particle points

• Δ = Inter‐particle spacing

– Δ =0.42 mm

• CSA =  pi * 2( σ2 + σ20)  

CT Assessment of Vascular Morphology

San Jose Estepar et al. AJRCCM, 2013

Carol
Typewritten Text
Quantification of Pulmonary Vasculature on CT Raúl San José Estépar, Ph.D. Brigham and Women's Hospital
Page 34: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Patient Population

COPDGeneCohort

(n=10,362)

COPDGeneCohort

(n=10,362)

First 2500 subjectsFirst 2500 subjectssubjectssubjects

NJH(n=359) NJH

(n=359) 

Never smokers Controls(n=82) 

Never smokers Controls(n=82) 

GOLD 0(n=104) GOLD 0(n=104) 

GOLD 1(n=27) GOLD 1(n=27) 

GOLD 2(n=90) GOLD 2(n=90) 

GOLD 3(n=87) GOLD 3(n=87) 

GOLD 4(n=51) GOLD 4(n=51) 

Population Characteristics

Never‐Smoking Controls (n=85)

Smokers              (n=359) P Value

Age 62.4 (56.3‐69.1) 63.7 (54.6 – 70.1) NS

FEV1pp 103.7 (93‐113) 59 (38 – 89) <0.0001

Height (cm) 163.2 (159.5‐171.6) 168.7 (163.0 – 176.1) NS

Weight (kg) 80.2 (63.8‐88.5) 78.6 (64.3 – 90.0) NS

BMI 27 5 (24 1 31 4) 26 65 (23 08 30 70) NSBMI 27.5 (24.1‐31.4) 26.65 (23.08 – 30.70) NS

%LAA‐950 0.8 (0.4 – 2.3) 6.33 (1.75 – 22.76) <0.0001

SO2 NA 93 (90 – 95)

6 MWD (ft) NA 1310 (1025 – 1650) SGRQ TotalScore

NA 32.16 (15.94 – 47.26)

BODE (n=352) NA 2 (0‐4)

DLCO (n=135) NA 13.11 (8.58 – 16.64)

CT ProtocolScanner ConfigurationScanner Make GE Siemens

Scanner ModelLight 

Speed 16Discovery HD750

DefinitionDefinition 

AS+Definition Flash

Scan Type Helical Helical Spiral Spiral Spiral

Rotation Time (s) 0.5 0.5 0.5 0.5 0.5

Det. Configuration 16x0.625 64x0.625 64x0.6 128x0.6 128x0.6

Pi h 1 375 1 375 1 1 1 1Pitch 1.375 1.375 1.1 1 1Speed (mm/rot) 13.75 13.75 13.2 38.4 38.4kVp 120 120 120 120 120

mAs 200 200 200 200 200Reconstruction

Algorithm Standard Standard B31f B31f B31f

Thickness (mm) 0.625 0.625 0.75 0.75 0.75Interval (mm) 0.625 0.625 0.5 0.5 0.5

SubjectsSmokers 42 317Non‐smokers 12 5 36 5 24

Blood Volume differences between Normal and Smokers

Dilation in the range 7-50 mm2

and pruning in less than 5 mm2G

OLD

0-2

GO

LD 3-4

BVCOPD < BVNormal

BVCOPD > BVNormal

Pruning

Dilation

Associations in Smokers

San Jose Estepar et al. AJRCCM, 2013

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Quantification of Pulmonary Vasculature on CT Raúl San José Estépar, Ph.D. Brigham and Women's Hospital
Page 35: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Therapy Assessment (I): ELVR

Slide provided by F. Rahahi

Endoscopic Lung Volume Reduction

ELVR Endoscopic Lung Volume Reduction 

25000

27000

29000

31000

33000

35000

30000

35000

40000

45000

14 patients in the prior Aeriseal trial with CT scans prior to ELVR, 12 weeks and 48 weeks.

15000

17000

19000

21000

23000

1 2 3

15000

20000

25000

1 2Week 0 Week 0Week 12 Week 12 Week 48

• +12% change between week 0 and week 12 (N = 13) and 22% change between week 0 and week 48 (N = 5). Consistent with hypothesis of decompression.

F Rahaghi, to be presented at ATS

Therapy Assessment (II): Nitric Oxide Challenge Nitric Oxide Challenge

Right Lung BV5mm2 BVTOT BV5mm2/BVTOT BV60‐90mm2

Baseline 7.63E+04 1.59E+05 47.87% 1.06E+04

+Nitric 8.41E+04 1.62E+05 51.82% 9.99E+03

Carol
Typewritten Text
Quantification of Pulmonary Vasculature on CT Raúl San José Estépar, Ph.D. Brigham and Women's Hospital
Page 36: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

What else can Vascular QCT offer?

Bronchiectasis• Assessment of Airway to Vessel Ratio (A/V) by matching 

closest points• Compute A/V distribution over vessel cross sectional area 

(CSA)

BronchiectaticNon-Bronchiectatic

• Early analyses show differences with between GOLD U and GOLD 0‐4 and being a independent prediction with severe exacerbations

> 10 mm2

> 10 mm2

Conclusions

• Pulmonary vascular morphology in smokers appears to be characterized by distal pruning and proximal dilation

• There is a discrete point in vessel size that characterizes that transition and can be detected by CT

• Difference are more significant in proximal vessels

• Future studies are necessary to fully characterize this transition point in COPD and other diseases

– Use for epidemologic investigations and therapeutic outcomes.

THANK YOU

Carol
Typewritten Text
Quantification of Pulmonary Vasculature on CT Raúl San José Estépar, Ph.D. Brigham and Women's Hospital
Page 37: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT in lung fibrosis

David A Lynch, MBDavid A Lynch, MB

Quantification of lung fibrosisQuantification of lung fibrosis

SemiquantitativeSemiquantitative Densitometry/CT histogramDensitometry/CT histogram

TextureTexture based methodsbased methods TextureTexture--based methodsbased methods

SemiquantitativeSemiquantitative assessment in assessment in IPFIPF

Extent of fibrotic abnormalityExtent of fibrotic abnormality Extent of ground glass abnormalityExtent of ground glass abnormality

Extent of emphysemaExtent of emphysema Extent of emphysemaExtent of emphysema 55--point scale: 0, 1point scale: 0, 1--25%, 2625%, 26--50%...50%... 1111--point scale: 0, 1point scale: 0, 1--10%, 1110%, 11--20%...20%... 2121--point scale: 0, 5%, 10%.....point scale: 0, 5%, 10%.....

Relationship between Relationship between semiquantitativesemiquantitativeassessment and physiologic impairmentassessment and physiologic impairment

Extent of IPF on CT was independently Extent of IPF on CT was independently related to:related to: % predicted DLCO r= 0.65, p <0.0005% predicted DLCO r= 0.65, p <0.0005 % predicted FVC% predicted FVC r= r= --0.53, 0.53, p <p <0.00050.0005 % predicted % predicted FEV1FEV1 r= r= --0.34, 0.34, p <p <0.020.02

Composite physiologic index:Composite physiologic index: extent of disease on CT = 91.0extent of disease on CT = 91.0-- (0.65 X (0.65 X

DLCO%)DLCO%)-- (0.53 X FVC%)+ (0.34 X FEV1(0.53 X FVC%)+ (0.34 X FEV1%)%)

Wells et al. Am J Respir Crit Care Med Vol 167. pp 962–969, 2003

Relationship between Relationship between semiquantitativesemiquantitative assessment and assessment and

physiologic impairmentphysiologic impairment

Wells et al. Am J Respir Crit Care Med Vol 167. pp 962–969, 2003

Relationship between semiquantitative assessment and mortality: Univariate

Baseline Variable Hazard ratio 95% Confidence Interval

p Value

Overall fibrosis extent

3.12 2.00, 4.89 < 0.0001

Reticulation pattern extent

2.69 1.71,4.23 < 0.0001

Honeycomb pattern extent

3.06 1.75, 5.34 < 0.0001extentPredominant pattern: reticulation

0.41 0.17, 0.99 0.04

Percent-predicted DLCO

0.92 0.89, 0.96 0.0001

A-a gradient 1.06 1.03, 1.09 < 0.0001Current oxygen use 2.37 1.29, 4.34 0.004

Percent-predicted FVC

0.97 0.94, 1.00 0.03

Lynch DA, et al. Am J Respir Crit Care Med2005;172:488-93.

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Carol
Typewritten Text
Carol
Typewritten Text
Page 38: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Relationship between semiquantitative assessment and mortality: Multivariate

Baseline Variable

Hazard ratio 95% Confidence Interval

p Value

HRCT features

Overall extent 2 71 1 61 4 55 0 0001Overall extent of fibrosis

2.71 1.61, 4.55 < 0.0001

% predicted DLCO

0.94 0.90, 0.98 0.004Treatment assignment to IFN-γ1b

0.53 0.28, 0.99 0.04

Lynch DA, et al. Am J Respir Crit Care Med2005;172:488-93.

Scleroderma lung studyScleroderma lung study

N= 158N= 158 Randomized cyclophosphamide Randomized cyclophosphamide vsvs placeboplacebo Mean absolute difference in FVC at 12 Mean absolute difference in FVC at 12

months= 2.53% (0.28, 4.9%), favoring months= 2.53% (0.28, 4.9%), favoring cyclophosphamide (p<0.03)cyclophosphamide (p<0.03)

When adjusted for worst zonal When adjusted for worst zonal semiquantitativesemiquantitative score for fibrosis on baseline score for fibrosis on baseline CT, mean absolute difference increased to CT, mean absolute difference increased to 2.97% (0.75, 5.19%), (P=0.0092.97% (0.75, 5.19%), (P=0.009))Tashkin et al. N Engl J Med 2006; 354:2655-2666

Limitations of Limitations of semiquantitativesemiquantitativescoringscoring

SubjectiveSubjective InterInter--observer variationobserver variation

Relative lack of precision for longitudinalRelative lack of precision for longitudinal Relative lack of precision for longitudinal Relative lack of precision for longitudinal analysisanalysis

CTCT--based quantification of lung based quantification of lung fibrosisfibrosis

Densitometry/CT histogramDensitometry/CT histogram Local histogram methodsLocal histogram methods

TextureTexture based methodsbased methods TextureTexture--based methodsbased methods

HistogramHistogram--based parametersbased parameters

Best et al., Radiology; 2003;208:407-414

HistogramHistogram--based parametersbased parameters

Mean lung attenuation Skewness

Degree of leftward or rightward deviation ofrightward deviation of histogram, compared with Gaussian

Kurtosis Degree of peakedness of

histogram, compared with Gaussian

Best et al., Radiology; 2003;208:407-414

Carol
Typewritten Text
Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 39: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Relationship between quantitative Relationship between quantitative histogram assessment and histogram assessment and

physiologic impairmentphysiologic impairment

R2= 0.23P<0.0001

R2= 0.30P<0.0001

R2= 0.20P<0.0001

Best et al., Radiology; 2003;208:407-414

UnivariateUnivariate analysis of predictors of analysis of predictors of mortality in IPF (n=167, 35 deaths)mortality in IPF (n=167, 35 deaths)

Baseline parameters Odds ratio estimates

95% Confidence Limits

Wald Chi-Square Pr > ChiSq

FVC (% predicted) 0.976 0.949 to 1.004 2.7323 0.0983

TLC (% predicted) 0.97 0.942 to 0.998 4.3191 0.0377

MLA (HU) 1 004 0 998 t 1 011 1 533 0 2157MLA (HU) 1.004 0.998 to 1.011 1.533 0.2157

Skewness 0.29 0.092 to 0.915 4.4584 0.0347

Kurtosis 0.503 0.287 to 0.883 5.7355 0.0166

CT extent of fibrosis (%) 1.125 1.041 to 1.217 8.714 0.0029

CT extent of GGO (%) 0.992 0.949 to 1.037 0.1328 0.7155

CT extent of emphysema (%) 0.986 0.894 to 1.088 0.0797 0.7777

Treatment assignment 0.86 0.311 to 2.378 0.0843 0.7716

Best et al., Radiology; 2008 Mar;246(3):935-40.

Multivariate analysis of predictors of Multivariate analysis of predictors of mortality in IPF (n=167, 35 deaths)mortality in IPF (n=167, 35 deaths)

Effect

Odds Ratio Estimates

95% Confidence Limits

Wald Chi-Square

Pr > ChiSq

Kurtosis at 0.32 to Baseline 0.579 1.049 3.249 0.0715

Mean Fibrosis at Baseline 1.104

1.018 to 1.198 5.7171 0.0168

Best et al., Radiology; 2008 Mar;246(3):935-40.

Changes in CT parameters over 12 Changes in CT parameters over 12 months (n=95)months (n=95)

Baseline 12 Months P value

Mean (SD) Mean (SD)

Mean lung attenuation (HU) -707.4 (60.8) -691.5 (62.4) 0.0032(HU) 707.4 (60.8) 691.5 (62.4) 0.0032

Skewness 1.043 (0.392) 0.911 (0.42) <0.0001

Kurtosis 0.497 (1.065) 0.204 (1.0) <0.0001

SemiquantitativeAssessment of fibrosis (%) 17.7 (5.4) 18.5 (6.0) 0.0302

Relationship between quantitative Relationship between quantitative and and semiquantitativesemiquantitative assessment assessment

and physiologic impairmentand physiologic impairment

Sverzellati et al. Radiol med (2007) 112:1160–1172

Sources of variation in histogram Sources of variation in histogram based analysisbased analysis

Inspiratory lung volumeInspiratory lung volume Careful instruction in breathing is mandatoryCareful instruction in breathing is mandatory

Associated emphysema Associated emphysema p yp y Technical factorsTechnical factors

CT doseCT dose Scanner reconstruction parametersScanner reconstruction parameters Scanner make/modelScanner make/model

Segmentation and analysis platformSegmentation and analysis platform

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 40: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Local histogram analysisLocal histogram analysis

Iwasawa et al J Thorac Imag 2009: 24: 216-222

Local histogram analysisLocal histogram analysis

Iwasawa et al J Thorac Imag 2009: 24: 216-222

Local histogram analysis Local histogram analysis vsvssurvivalsurvival

Iwasawa et al J Thorac Imag 2009: 24: 216-222

Dissimilarity assessment of Dissimilarity assessment of cumulative density functionscumulative density functions

Bartholmai et al. Journal of Thoracic Imaging. 2013;28(5):298-307

GlyphGlyph--based characterization of based characterization of lung parenchymalung parenchyma

Bartholmai et al. Journal of Thoracic Imaging. 2013;28(5):298-307

Texture based methodsTexture based methods

Adaptive multiple feature methodAdaptive multiple feature method Quantitative lung fibrosis scoreQuantitative lung fibrosis score

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 41: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Adaptive multiple feature methodAdaptive multiple feature methodAMFMAMFM

Includes Includes FirstFirst--order histogram featuresorder histogram features RunRun--length, and colength, and co--occurrence featuresoccurrence features Fractal featuresFractal features Fractal featuresFractal features

Identify and visually characterize features of Identify and visually characterize features of interest (ground glass, reticular, honeycombing, interest (ground glass, reticular, honeycombing, etc.)etc.)

Optimal feature selection using Bayesian Optimal feature selection using Bayesian classificationclassification

Apply classification to test setApply classification to test setXu Y, et al. Acad Radiol. 2006;13(8):969-78.

Adaptive multiple feature Adaptive multiple feature methodmethod

Honey-comb

GGO Broncho-vascular

Nodular Emphy-sema

Normal

Sensitivity 82.5 94.3 89.2 86.7 99.2 92.7Sensitivity (%)

82.5 94.3 89.2 86.7 99.2 92.7

Specificity (%)

99.9 99.1 98.9 96.6 100 97.8

Uppaluri R et al. Am J Respir Crit Care Med. 1999 Aug;160(2):648-54.

Adaptive multiple feature method:Adaptive multiple feature method:2d2d--3d3d

AMFM Whole lung classificationAMFM Whole lung classification

Honeycomb

Ground Glass –Reticular

B h l

Ground Glass

Bronchovascular

Normal

Emphysema

Flaherty, ATS 2013

Quantitative lung fibrosis (QLF) Quantitative lung fibrosis (QLF) scorescore

Kim et al. Clin Exp Rheumatol. 2010 ; 28(5 Suppl 62): S26–S35.

Quantitative lung fibrosis (QLF) Quantitative lung fibrosis (QLF) scorescore

Kim et al. Clin Exp Rheumatol. 2010 ; 28(5 Suppl 62): S26–S35.

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 42: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scleroderma: sequential Scleroderma: sequential evaluationevaluation

Kim et al, Eur Radiol. 2011 Dec;21(12):2455-65

Scleroderma: sequential Scleroderma: sequential evaluationevaluation

Kim et al, Eur Radiol. 2011 Dec;21(12):2455-65

Textural analysis at NJH: Textural analysis at NJH: TrainingTraining

Expert selection of representative ROIs (confirmed by Expert selection of representative ROIs (confirmed by 2nd) demonstrating characteristic patterns (2nd) demonstrating characteristic patterns (normal normal and and IPF) in cases drawn from established IPF) in cases drawn from established IPFNetIPFNet studystudy

Retain full DICOM metadata to control imaging Retain full DICOM metadata to control imaging parametersparameters

Feature extractionFeature extraction

•• 15 x 15 pixel ROIs from radiologist drawn patches15 x 15 pixel ROIs from radiologist drawn patches•• Compute variety of local features:Compute variety of local features:

-- Kernel Density Estimate (KDE) of local intensity histogramKernel Density Estimate (KDE) of local intensity histogram-- Local Binary Pattern (LBP)Local Binary Pattern (LBP)

Hi t f O i t d G di t (HOG)Hi t f O i t d G di t (HOG)-- Histogram of Oriented Gradients (HOG)Histogram of Oriented Gradients (HOG)-- Scale Invariant Feature Transform (SIFT)Scale Invariant Feature Transform (SIFT)

ROI Label

Fibrotic Non-fibrotic

Classifier prediction

Fibrotic 11257 202

Non fibrotic 179 3026

Textural analysis at NJH: results summary

Non-fibrotic 179 3026

Sensitivity 11257 / (11257+179) 98.4%

Specificity 3026 / (3026+202) 93.7%

Pos.predicted 11257 / (11257+202) 98.2%

Neg.predicted 3026 / (179+3026) 94.4%

NJH: Textural analysisNormal lung Reticular

Abnormality

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 43: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

NJH: Textural analysisNJH: Textural analysisNormal lung

Reticular Abnormality

Advantages of imageAdvantages of image--based based quantification in IPFquantification in IPF

Provides morphologic characterizationProvides morphologic characterization ReticularReticular HoneycombHoneycomb Ground glassGround glass

I l d f i l i f iI l d f i l i f i Includes functional informationIncludes functional information Lung volumesLung volumes

Provides regional and lobar informationProvides regional and lobar information Sensitive, and increasingly sophisticated, quantitative Sensitive, and increasingly sophisticated, quantitative

tools are availabletools are available Emerging information on reproducibility, longitudinal Emerging information on reproducibility, longitudinal

change and normal valueschange and normal values

Textural analysis in IPF: limitationsTextural analysis in IPF: limitations

Sources of variationSources of variation Level of inspirationLevel of inspiration Scanner make/model/reconstruction algorithmScanner make/model/reconstruction algorithm Scanner make/model/reconstruction algorithmScanner make/model/reconstruction algorithm CT doseCT dose Analysis techniqueAnalysis technique

Radiation doseRadiation dose More validation is neededMore validation is needed

SummarySummary

Textural quantitative imaging has Textural quantitative imaging has substantial potential value for substantial potential value for quantification and characterization of quantification and characterization of qqsarcoid and IPFsarcoid and IPF

Carol
Typewritten Text
Quantitative CT in Lung Fibrosis David A. Lynch, MD National Jewish Health
Page 44: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Carol
Typewritten Text
Page 45: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Carol
Typewritten Text
Carol
Typewritten Text
Page 46: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 47: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Page 48: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Page 49: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Page 50: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM
Carol
Typewritten Text
Carol
Typewritten Text
Texture-Based Approaches to Image Analysis Joyce D. Schroeder, MD National Jewish Health
Carol
Typewritten Text
Carol
Typewritten Text
Page 51: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

2014.03.15. Society of Thoracic Radiology 2014 QCT Imaging of the Lung (San Antonio, TX, USA)

          

MRI of the Lung      

 

Yoshiharu Ohno, M.D., Ph.D. 1, 2  1 Advanced Biomedical Imaging Research Center, Kobe University Graduate

 

School of Medicine  2Division of Functional Imaging Research, Department of Radiology, Kobe

 

University School of Medicine

Objective  

 

1. To learn the state-of-the art lung MR imaging techniques

for various chest disease.

2. To understand the potential of quantitatively assessed    

MR imaging for various chest diseases.

3. To learn the clinical utility of quantitative and semi-

quantitative assessment of lung MR imaging for not only

diagnosis, but also patients’ management and treatment

response in various pulmonary diseases.

Aims for Quantitative Assessment on Lung MRI

 

1. Physiology and Physiopathology Assessment in normal and Pulmonary Disease Subjects

2. Diagnosis  

• Pulmonary Nodule and Mediastinal Tumor  

• N-Staging of Lung Cancer  

3. Disease Severity Assessment  

• Pulmonary Vascular Diseases  

• Airway and Lung Parenchyma Diseases  

• Thoracic Malignancies  

4. Patients’ Care and/ or Cure  

• Thoracic Malignancies  

• Pulmonary Vascular Diseases  

• Airway and Lung Parenchyma Diseases

Quantitative Methods for Lung MRI  

1. Oncology:  

• STIR Turbo SE Imaging  

• Diffusion-Weighted MR Imaging  

• Dynamic Contrast-Enhanced First-Pass Perfusion MR Imaging  

• Oxygen-Enhanced MR Imaging  

2. Pulmonary Vascular Diseases  

• Dynamic Contrast-Enhanced First-Pass Perfusion MR Imaging  

• Phase-Contrast MR Imaging  

• Hyper-Polarized Noble Gas MR Imaging (and Oxygen-Enhanced  

MR Imaging)  

3. Airway and Lung Parenchyma Diseases  

• Hyperpolarized Noble Gas MR Imaging  

• Oxygen-Enhanced MR Imaging  

• Dynamic Contrast-Enhanced First-Pass Perfusion MR Imaging  

• Pulmonary MR Imaging with Ultra-Short TEs

STIR Turbo SE Imaging  

75-year-old male patient with metastatic LNs from invasive Ad.  

                 

CE-CT STIR turbo SE Imaging PET/CT   

67-year-old female patient with non-metastatic LN from invasive Ad.                  

CE-CT STIR turbo SE Imaging PET/CT

STIR Turbo SE Imaging vs. CE-CT      

Contrast- Quantitatively assessed Qualitatively assessed enhanced CT STIR images STIR images

 

    

Sensitivity (%) 53 93* 88*

  

 

(21/40) (37/40) (35/40)   

    

Specificity (%) 83 87 86

  

 

(58/70) (61/70) (60/70)   

    

Accuracy (%) 72 89* 86*

  

 

(79/110) (98/110) (95/110)    

 

Ohno Y, et al. Radiology. 2004; 231: 872-879.

Carol
Typewritten Text
Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 52: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

STIR Turbo SE Imaging vs. FDG-PET/CT  

     

Modality Index Feasible TV SE (%)  

    

90.1

SP (%)  

    

93.1

AC (%)  

    

92.2  

STIR turbo SE imaging  

LSR  

0.6     

(39/43)

   

(67/72)

    

(106/115)

    

 

Co-registered FDG-PET/CT

    

 

SUV

    

 

2.0

  

76.7*  

    

(33/43)

 

87.5  

    

(63/72)

  

83.5*  

    

(96/115) 

 

TV: Threshold value, SE: Sensitivity, SP: Specificity, AC: Accuracy  

LSR: Lymph node-salilne ratio  

*: Signifiacant difference with LSR (p<0.05)   

Ohno Y, et al. J Magn Reson Imaging. 2007; 26: 1071-1780.

Diffusion-Weighted MR Imaging (DWI)  

75-year-old female patient with invasive adenocarcinoma (N1)                   

Thin-Section CT STIR turbo SE Imaging DWI    

73-year-old female patient with invasive adenocarcinoma (N1)  

              

Thin-Section CT STIR turbo SE Imaging DWI PET/CT

Diffusion-Weighted MR Imaging (DWI)     

SE (%) SP (%) AC (%)         

STIR turbo SE imaging

  

LSR  

     

LMR

82.7 89.2 86.8

(77/93) (140/157) (217/250)

82.7 89.2 86.8   

(77/93) (140/157) (217/250)      

DWI ADC (×10-3s/mm2) 74.2*, ** 90.4 84.4*, **    

(69/93) (142/157) (211/250)     

FDG-PET/CT SUVmax

74.2*, ** 94.9 85.6   

(69/93) (145/157) (214/250)  LSR: Lymph node-salilne ratio, LMR: Lymph node-muscle ratio *: Signifiacant difference with LSR (p<0.05) **: Signifiacant difference with LMR (p<0.05)  

Ohno Y, et al. Radiology. 2011; 261: 605-615.

N-Staging in Non-Small Cell Lung Cancer  

-STIR Turbo SE Imaging and DWI-   

• Quantitatively assessed STIR turbo SE imaging has better

capability for diagnosis of lymph node metastasis than contrast-

enhanced CT, FDG-PET or PET/CT and DWI in non-small cell lung

cancer (NSCLC) patients.

• Quantitative Assessment of DWI using ADC has equal capability

for N-staging in NSCLC patients as compared with FDG-PET/CT.

• Further DWI sequence improvement and analysis software

development will provide better potential, and encourage this

technique in routine clinical practice.   

Ohno Y, et al. Radiology. 2004; 231: 872-879. Ohno Y, et al. J Magn Reson Imaging. 2007; 26: 1071-1780. Ohno Y, et al. Radiology. 2011; 261: 605-615.

Dynamic MR Imaging for Nodule Management  

-SE vs. Ultra-Fast GRE (or Fast GRE)-  

   

SE (or Turbo SE) technique Ultra f-GRE (or f-GRE) technique           

 

t=0s t=150s t=8.8s t=9.9s t=15.4s  

 

Adenocarcioma Adenocarcioma  

           

t=0s t=150s t=0s t=12.1s t=23.1s  

Tuberculoma  

Tuberculoma

Dynamic CT vs. PET/CT vs. Dynamic MRI -Differentiation of Malignant Nodules from Benign Nodules-

 

  

  

Maximum enhancement combined with

absolute loss of enhancement  

Net enhancement combined with absolute  

loss of enhancement

Cut-off value SE (%) SP (%) AC (%)  

93.4 42.0 a, b, c 80.7 a, b, c 60.0HU and ≤30HU

(142/152) (21/50) (163/202)  

94.7 52.0b 83.2b 20.0 HU and ≤30 HU

(142/152) (26/50) (168/202)  

Slope of enhancement combined with

absolute loss of enhancement

0.4HU/sec and  

≤30HU

94.7 48.0b 82.2a, b  

(142/152) (24/50) (166/202)  

96.0 54.0 85.6 Maximum relative enhancement ratio 0.2

   

Slope of enhancement ratio 0.04/sec    

SUVmax 1.8

 

(146/152) (27/50) (173/202)  

96.0.0 64.0 88.1 (146/152) (32/50) (178/202)

93.4 54.0 83.6b  

(142/152) (27/50) (169/202)  

SE: Sensitivity, SP: Specificity, AC: Accuracy a: Significant difference from maximum relative enhancement ratio (p<0.05). b: Significant difference from slope of enhancement ratio (p<0.05). c: Significant difference from SUVmax (p<0.05).

 

 

Ohno Y, et al. J Magn Reson Imaging. 2008; 27: 1284-1295.

Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 53: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Dynamic MR Imaging for Nodule Management 

-SE vs. Ultra-Fast GRE (or Fast GRE)-   

• Dynamic CE-MR imaging has equal potential for distinguishing

malignant from benign nodules, when compared with dynamic CT,

FDG-PET and 99mTc-depreotide SPECT.

• Dynamic CE-MR imaging has performed by various sequences and

techniques, and drawbacks are time consuming or low spatial

resolution.

• Quantitative software as well as sequence improvements and

technical reproducibility will encourage this technique in routine

clinical practice.   

Ohno Y, et al.Radiology. 2002; 224: 503-511. Ohno Y, et al. J Magn Reson Imaging. 2008; 27: 1284-1295. Cronin P, et al. Radiology. 2008; 246: 772-782.

Quantitatively Assessed Regional Perfusion Parameters using Dynamic

CE-First Pass Perfusion MRI                          

Pulmonary Blood Flow (PBF) Map

Pulmonary Blood Volume (PBV) Map

Mean Transit Time (MTT) Map

63-year-old male subject with adenocarcinoma in the left upper lobe            

Quantitative CT          

Perfusion SPECT  

                   

        

Dynamic perfusion MRI

Ohno Y, et al.AJR Am J Roentgenol. 2007; 189: 400-408.

Prediction of Postoperative Lung Function  

-Dynamic perfusion MRI vs. Quantitative and Qualitative CT vs. Perfusion SPECT-

The limits of agreement = 5.3±11.8 % The limits of agreement = 5.0±11.6 %                  

The limits of agreement = 6.8±14.4 % The limits of agreement = 5.1±14.0 %                     

Ohno Y, et al.AJR Am J Roentgenol. 2007; 189: 400-408.

66-year-old male with acute pulmonary embolism.  

          

Contrast-enhanced MDCT  

            

Time-Resolved CE-MR Angiography                 

PBF map generated from dynamic CE-perfusion MRI  

Ohno Y, et al. J Magn Reson Imaging. 2010; 31: 1081-1090.

Capability for Distinguishing Death from Survival Groups in ACPTE Patients

-CE-CT vs. Time-Resolved CE-MRA vs. CE-Perfusion MRI-  

 

RV/LV diameter ratio PECT index PEMRA index PEPerfusion MRI index  

  

Az 0.96 0.86*, ** 0.86*, ** 0.96  

Feasible threshold value 1.6≤ 60% ≤ 60% ≤ 60% ≤

 

62.5 62.5 62.5 62.5 Sensitivity (%)  

(5/8) (5/8) (5/8) (5/8)   

 

Specificity (%) 97.6 83.3*, ** 83.3*, ** 97.6

 

(41/42) (35/42) (35/42) (41/42)   

 

Accuracy (%) 92.0 80.0 80.0 92.0

(46/50) (40/50)*, ** (40/50)*,** (46/50)

AZ: Area under the curve RV: right ventricle; LV: left ventricle *: Significant difference with RV/LV diameter ratio (p<0.05). **: Significant difference with PEPerfusion MRI index (p<0.05).

Ohno Y, et al. J Magn Reson Imaging. 2010; 31: 1081-1090.

Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 54: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Capability as Survival Predictor in ACPTE Patients  

 

-CE-CT vs. Time-Resolved CE-MRA vs. CE-Perfusion MRI-      

   Odds ratio* 95% CI p value 

RV/LV diameter ratio  

1.17  

1.05-1.31 0.004  

PECT index  

1.10  

1.03-1.18 0.004  

PEMRA index  

1.16  

1.05-1.27 0.004  

PEPerfusion MRI index  

1.21  

1.06-1.37 0.004           

 

Ohno Y, et al. J Magn Reson Imaging. 2010; 31: 1081-1090.

Quantitative CE-Perfusion MRI 42-year old female primary pulmonary hypertension (PPH) patient

 

                                  

Quantitative PBF Quantitative PBV MTT       

Ohno Y, et al. AJR Am J Roentgenol. 2007; 188: 48-56

Quantitative Dynamic First-Pass CE-Perfusion MRI -Disease Severity Assessments in PAH-

 

          

r=-0.79, r2=0.62, p<0.001

r=0.60, r2=0.36, p=0.022

 

                       

Ohno Y, et al. AJR Am J Roentgenol. 2007; 188: 48-56 Ohno Y, et al. J Magn Reson Imaging. 2012; 36: 612-623.  

76-year-old male with chronic thromboembolic pulmonary hypertension treated with conservative therapy and assessed as responder

Ohno Y, et al. J Magn Reson Imaging. 2012; 36: 612-623. 

Comparison of Physiological, CT- and MR-based indexes between CPTEH patients assessed as

responder and non-responders -CE-CT vs. Time-resolved MRA vs. Perfusion MRI-

 

 

  Responder group (n=13)

Mean+/-SD

0.2±0. 1  

86.5±52.9  

-9.9±10.5  

-192.8±161.2  

-0.3±0.1  

-13.1±8.7  

-17.8±10.4  

33.4±18.2  

4.5±2.7  

-1.3±1.3

Non-responder group (n=11)

Mean+/-SD

-0.1±0.2*  

-16.0±71.3*  

2.8±6.5*  

31.7±63.5*  

0.2±0.3*  

0.7±3.7*  

0.3±4.2*  

0.6±8.8*  

0.3±1.9*  

0.1±0.5*

 

P value

   

Ohno Y, et al. J Magn Reson Imaging. 2012; 36: 612-623.  

Improvement in:      

CI (l/min/mm2)     <0.0001

6-MWD (m)     0.0006

mPAP (mmHg)     0.0023

PVR (dyne/s-1/cm-5)     0.0003  

RV/LV diameter ratio      

<0.0001  

CTEPHCTA (%)      

<0.0001

CTEPHMRA (%)     <0.0001

PBF (ml/100ml/min)     <0.0001

PBV (ml/100ml)     0.0004

MTT (sec)     0.0035 *: Significant difference with responder group (p<0.05).

Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 55: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Ohno Y, et al. J Magn Reson Imaging. 2012; 36: 612-623. 

SE (%)  

100

SP (%)  

81.8

AC (%)  

91.7

(13/13) (9/11) (22/24)

100 36.4 70.8

(13/13) (4/11) (17/24)

100 54.5 79.2

(13/13) (6/11) (19/24)

100 90.9* 95.8*

(13/13) (10/11) (23/24)

100 72.7 87.5

(13/13) (8/11) (21/24)

100 72.7 87.5

(13/13) (8/11) (21/24)

Comparison of CT- and MR-based indexes between CPTEH patients assessed as responder and non-responders

-CE-CT vs. Time-resolved MRA vs. Perfusion MRI-   

Improvement in: Feasible threshold value  

  

RV/LV diameter ratio ≤0  

   

CTEPHCTA (%) ≤0   

 

CTEPHMRA (%) ≤0   

 

PBF (ml/100ml/min) 7≤  

   

PBV (ml/100ml) 0.9<  

   

MTT (s) <1    

*: Significant difference with CTEPHCTA (p<0.05).

 

Quantitatively Assessed Dynamic CE-First- Pass Perfusion MRI for Chest Disease

• Dynamic CE-first-pass perfusion MR imaging is able to assess regional perfusion changes due to pulmonary emphysema and predict postoperative lung function as well as thin-section CT and perfusion SPECT in NSCLC patients.

• Dynamic CE-first-pass perfusion MR imaging has the potential to assess physiopathology and disease severity in PVD patient as non- invasive method as compared with right heart catheterization.

• Dynamic CE-first-pass perfusion MR imaging can be play as one of the biomarkers for treatment and management as well as one of the predictors for patient outcome in PVD patients.

• Quantitative software as well as sequence improvements and technical reproducibility will encourage this technique in routine clinical practice.

 

 

Ohno Y, et al. J Magn Reson Imaging. 2004;20: 353-65. Ohno Y, et al. AJR Am J Roentgenol. 2007; 188: 48-56. Ohno Y, et al. J Magn Reson Imaging. 2010; 31(5): 1081-1090. Ohno Y, et al. J Magn Reson Imaging. 2012; 36: 612-623.

 

Phase-Contrast MRI

 

Time-velocity course curves of normal volunteer and patient with high pulmonary

vascular resistance (PVR)  

Peak velocity (PV)

                  

 

Acceleration time (AT)

Nogami M, et al. J Magn Reson Imaging. 2009 ; 30: 973-980. . 

Measurement Error Evaluation of SV and PASP -Cardiac US vs. Phase-Contrast MRI-

 

 

Between PC-MRI  

and catheterization

Between cardiac US  

and catheterization        

SV [ml]

(n=11)

Mean difference 1.0 9.0

(95%CI of mean difference) (-1.4 ~ 8.4) (-6.0 ~ 56.6)

Upper limits of agreement 7.7 52.2

(95% CI of upper limits of agreement) (3.7 ~ 11.8) (26.2 ~ 78.3)

Lower limits of agreement -5.8 -34.1

(95 %CI of lower limits of agreement) (-9.8 ~ -1.7) (-60.2 ~ -8.1)  

Mean difference -3.0 -6.5     

PASP

[mmHg]

(n=11)

(95%CI of mean difference) (-7.6 ~ 11.8) (-16.5 ~ 25.0)

Upper limits of agreement 10.5 22.2

(95% CI of upper limits of agreement) (2.3 ~ 18.6) (4.9 ~ 39.5)

Lower limits of agreement -16.5 -35.2

(95 %CI of lower limits of agreement) (-24.6 ~ -8.3) (-52.5 ~ -17.9)

Alunni JP, et al. Eur Radiol. 2010; 20: 1149-1159.  

PAH Assessment on Phase-Contrast MRI  

-Capability for Disease Severity Assessments-    

PH grades    

Grade I  

Grade II  

Grade III  

Grade IV  

p value  

RV Ejection Fraction  

36.5  

32  

21*  

13**  

0.0005  

RV Ejection Systolic Volume  

71.5  

66.9  

119.6*  

141.7**  

0.0059  

RV Area Change  

31  

33.5  

19.5*  

10.5**  

0.0010  

Systolic RV/LV Area Ratio  

1.3  

1.5  

2.4  

4.0**  

0.007  

Diastolic RV/LV Area Radtio  

1  

0.8  

1.6  

2.4**, ***  

0.0005   

*: Significant difference between grade I and III (p<0.05). **: Significant difference between grade I and IV (p<0.05). ***: Significant difference between grade III and IV (p<0.05).

Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 56: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

 

Ventilation MR Imaging 1. Hyperpolarized Noble Gas MR Imaging (HP-MRI)

 

• Qualitative and quantitative assessment of gas movement in airway and alveoli

• Difficulty for clinical installation due to gas and equipment issues

• Available in the limited western countries, and limited improvements of problematic factor for clinical set in near future.

2. Oxygen-Enhanced MR Imaging (O2-Enhanced MRI) • Qualitative and quantitative assessment of oxygen ventilation,

diffusion and uptake due to respiration • Difficulty for clinical installation due to MR sequence and

software issues, but no difficulty about equipment issue • Available in all over the world, and easier improvement of

problematic factors for clinical set  

Hyperpolarized Noble Gas MRI (HP-MRI)  

1. 3He: • Low solubility • High SNR due to high gyromagnetic ratio • High general polarization level (≥40%). • Cost: ≥ US$100/L • More limited availability in the clinical respiratory medicine

than 129Xe 2. 129Xe:

• Naturally abundant • Low SNR due to low gyromagnetic ratio • Low general polarization level (≤20%) • Cost:

Natural Xe gas (26%): ≤ US$20/L Higher concentration Xe gas (80%): ≥ US$700/L

• Anesthesia side effect • Higher clinical availability than 3He, if some problematic

factors solved

 

Oxygen-Enhanced MRI  

 O2-Enhanced MRI in Normal Subject O2-Enhanced MRI in COPD Subject

                           

• Functional Assessment based on not only Ventilation, but also Oxygen  

Diffusion in Smokers, COPD, Asthma and Lung Cancer.  

• Potential for Directly Assessment of Respiration with High Spatial  

Resolution and without Ionizing Radiation.  

• No need for Special Equipment and Gas

 

Oxygen-Enhanced MRI vs. Quantitative CT -Smoking-Related COPD-

      

Cases  

Age (years)

Smokers without COPD

 

40  

62 ± 13

Mild COPD  

40  

62 ± 12

Moderate COPD

 

40  

62 ± 11

Severe or Very Severe COPD  

40  

62 ± 10  

Lifetime smoking exposure (pack years) 32 ± 27 35±23.7 62±22a, b 79 ± 43a, b  

FEV1/FVC% (%)   

%FEV1 (%)

85.4 ± 5.0   

92.0 ± 10.2

65.0 ± 3.9a   

84.8 ± 4.9a

51.1 ± 9.6a, b   

62.4 ± 8.7a, b

37.3 ± 12.0a, b, c   

37.7 ± 7.1a, b, c  

 

%DLCO/VA (%) 90.4 ± 18.3 78.5 ± 13.8a 65.0 ± 15.0a, b 47.8 ± 13.6a, b, c   

O2-enhanced MRI parameter 0.21 ± 0.07 0.16 ± 0.04a 0.13 ± 0.04a, b 0.09 ± 0.04a, b, c   

Quantitative CT parameter 0.70 ± 0.14 0.61 ± 0.13 0.50 ± 0.16a, b 0.43 ± 0.18a, b  

 

a: Significant difference with ‘Smokers without COPD’ group (p<0.05). b: Significant difference with ‘Mild COPD’ group (p<0.05). c: Significant difference with ‘Moderate COPD’ group (p<0.05).

 

  

Ohno Y, et al. Am J Respir Crit Care Med. 2008; 177: 1095-1102.

 

42-year-old male candidate for lung volume reduction surgery  

            

Quantitative Thin-Section CT Perfusion SPECT/CT              

O2-enhanced MRI before LVRS                

 

O2-enhanced MRI after LVRS Ohno Y, et al. AJR Am J Roentgenol. 2012; 199: 794-802.

 

Correlations between Improvements in Oxygen-Enhancement  

before and after LVRS and in Clinical Outcomes.  

 

r=0.71, p<0.0001

r=0.64, p<0.0001

 

                  

r=0.45, p=0.02

r=0.44, P=0.02

 

              

Ohno Y, et al. AJR Am J Roentgenol. 2012; 199: 794-802.

Carol
Typewritten Text
Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Carol
Typewritten Text
Carol
Typewritten Text
Page 57: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

 

76-year-old smoking subject with Brinkman’s index of 1480  

                    

Mean wash-in time = 53.0 sec Mean relative enhancement ratio = 0.08                          

Ohno Y, et al. AJR Am J Roentgenol. 2008; 190: W93-W99.  

Dynamic O2-Enhanced MR Parameters vs.

Pulmonary Function Test Results  

r=0.80, r2=0.64,

p<0.0001

r=0.72, r2=0.52,

p<0.0001

r=0.72, r2=0.52,

p<0.0001  

            

r=0.69, r2=0.48,

p<0.0001

r=0.78, r2=0.61,

p<0.0001

r=0.62, r2=0.38,

p<0.0001            

 

Ohno Y, et al. AJR Am J Roentgenol. 2008; 190: W93-W99.

 

Quantitatively Assessed Ventilation Imaging by HP-MRI and O2-Enhanced MRI

• Ventilation imaging consisted with HP-MRI and O2-enhanced MRI is currently applied for academic purposes rather than clinical purposes due to some problematic factors that have to be solved for clinical set.

• O2-enhanced MR imaging has the potential to assess physiopathology and disease severity in COPD, asthma and interstitial lung disease due to connective tissue disease patients.

• O2-enhanced MR imaging can be play as one of the biomarkers for treatment and management as well as one of the predictors for patient outcome in pulmonary diseases.

• Quantitative software as well as sequence improvements and technical reproducibility will encourage O2-enhanced MR imaging in routine clinical practice.

Ohno Y, et al. Am J Respir Crit Care Med. 2008; 177: 1095-1102 Ohno Y, et al. AJR Am J Roentgenol. 2008; 190: W93-W99. Ohno Y, et al. Eur J Radiol. 2011 Jan;77(1):85-91. Ohno Y, et al. AJR Am J Roentgenol. 2012; 199: 794-802. Ohno Y, et al. Eur J Radiol. 2013 Nov 23 [Epub ahead of print]

 

Lung MRI with Ultra-Short TEs in Smokers on 3T System                  

Smokers wth Mild COPD Smokers wth Severe or very Severe COPD             

CT-based FLV was 0.72, and WA% was 60 %             

Mean T2* value for this subject was 0.75 msec.

CT-based FLV was 0.63, and WA% was 65 %              

Mean T2* value for this subject was 0.67 msec.   

Ohno Y, et al. AJR Am J Roentgenol. 2011; 197: W279-285.

 

Comparison of Capability for Difference AssessmentAmong All Clinical Stage in Smokers

-Quantitative CT vs. UTE-MRI-   

Smokers  

 

without COPD

Smokers with  

 

mild COPD

Smokers with  

 

moderate COPD

Smokers with severe or  

 

very severe COPD  

 

(Mean ± SD) (Mean ± SD) (Mean ± SD) (Mean ± SD)   

 

CT-based FLV (%) 76.3 ± 5.8 73.6 ± 6.7 60.1 ± 8.5 a, b 51.8 ± 10.0 a, b      

WA% (%) 63.1 ± 4.0 64.3 ± 4.9 66.6 ± 4.8 75.5 ± 2.9 a, b, c     

Mean T2* value (ms) 0.82 ± 0.09 0.74 ± 0.07 0.61 ± 0.07 a, b 0.46 ± 0.11 a, b, c   

SD: Standard deviation; CT-based FLV: CT-based functional lung volume; WA%: Ratio of WA to total airway area; T2*: T2 Star; a: Significant difference with smokers without COPD (p<0.05). b: Significant difference with smokers with mild COPD (p<0.05). c: Significant difference with smokers with moderate COPD (p<0.05).

 

Ohno Y, et al. AJR Am J Roentgenol. 2011; 197: W279-285.  

Lung MRI with Ultra-Short TEs in CTD Patients on 3T System              

53-year-old male assessed as normal subject  

             

46-year-old female patient with mild interstitial lung disease due to PSS              

65-year-old female patient with mild interstitial lung disease due to MCTD  

Ohno Y, et al. Eur J Radiol. 2013; 82: 1359-1365.

Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 58: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

 

Correlations of mean T2* value and CT-based disease severity with pulmonary functional test results and serum KL-6 for CTD patients only

-Quantitative CT vs. UTE-MRI-  

  

Mean T2* Value CT-Based Disease Severity    

  r p value r p value 

 

%FEV1

 

 

-0.06  

 

0.86  

 

-0.15  

 

0.56

 

%VC  

-0.53  

0.02  

-0.26  

0.31  

 

%DLCO

 

 

-0.63  

 

0.0049  

 

-0.53  

 

0.02

 

Serum KL-6  

0.52  

0.03  

0.44  

0.07   

 

Ohno Y, et al. Eur J Radiol. 2013; 82: 1359-1365.  

Quantitative Morphological Change  

Assessment by MRI with UTEs (UTE-MRI)  

• Regional T2* measurement on MRI with UTEs (UTE-MRI) has a potential to quantitative assessment of morphological changes within the lungs due to smoking-related COPD and interstitial lung disease due to connective tissue disease, and may play as complementary role for management of these patients with quantitatively and qualitatively assessed thin-section CT.

• UTE-MRI is one of the new techniques for pulmonary functional MR imaging, and 3 Tesla system may have an advantages in this issue, although further investigations are warranted.

• Quantitative software as well as sequence improvements and technical reproducibility will encourage this technique in routine clinical practice.

Takahashi M, et al. J Magn Reson Imaging. 2010 ; 32: 326-33. Ohno Y, et al. AJR Am J Roentgenol. 2011; 197: W279-285. Ohno Y, et al. Eur J Radiol. 2013; 82: 1359-1365. Ohno Y, et al. J Magn Reson Imaging. 2013 Sep 30 [Epub ahead of print]

 

Conclusion   

1. Quantitative methods in lung MRI can be applied not only

physiologic or physiopathologic assessments, but also evaluating

various pulmonary diseases.

2. Several techniques are ready as clinical applications, and others

will be clinically set in near future.

3. These techniques will be shifting from 1.5T to 3.0T MR systems as

clinical tools with developing new techniques and software in

near future.  

 

4. Softwares for quantitative analyses as well as MR system,

sequence and reconstruction algorithm developments are key

issues for quantitative assessment for lung MRI.

Carol
Typewritten Text
Carol
Typewritten Text
MRI of the Lung Yoshiharu Ohno, MD, PhD Kobe University School of Medicine
Page 59: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Establishing a “Normal” Baseline in QCT

Eric A. Hoffman, Ph.D. Dept. of Radiology, University of Iowa Carver College of Medicine

Advanced Pulmonary Physiomic Imaging Laboratory - APPIL

Disclosure: EAH is a founder and shareholder in VIDA Diagnostics

Hoffman EA. JAP 59:468-480, 1985

Standardized Volumetric Lung Imaging Includes Coaching to a Standardized Lung Volume

Full

Lung

Sca

n

Density Threshold for Emphysema: -950/910 HU

Calculate % of voxels below -950 or -910 HU in each lung region

Air Trapping Density Threshold: <-856HU

Air Trapping vs. dAge

p-value < 0.0001

Air Trapping vs. Location

p-value < 0.0001

Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Carol
Typewritten Text
Page 60: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Air Trapping vs. Location (by dAge)

p-value < 0.0001

Air Trapping vs. Gender (by dAge)

p-value < 0.0001

6 Standardized Airway Paths

RB1 RB4 RB10

LB1 LB4 LB10

Female vs. Male Airway Wall Thickness 56M 76F Normal Non-Smokers

Branching Variations

Total (n=222)

Male (n=103)

Female (n=119)

Right upper trifurcation (B1, B2, B3) 51.8% 59.2% 45.4% bifurcation (B1+B2, B3) 15.8% 15.5% 16.0% bifurcation (B1+B3, B2) 14.9% 12.6% 16.8% bifurcation (B1, B2+B3) 17.6% 12.6% 21.8%

Right lower bifurcation (B8, B9+B10) 89.6% 91.3% 88.2% bifurcation (B8+B9, B10) 6.3% 3.9% 8.4% trifurcation (B8, B9, B10) 4.1% 4.9% 3.4%

Left upper bifurcation (superior divison, lingular) - bifurcation (B1+2, B3) 87.8% 86.4% 89.1% bifurcation (superior divison, lingular) - trifurcation 8.6% 8.7% 8.4% trifurcation (B1+2, B3, lingular) 3.6% 4.9% 2.5%

Left lower bifurcation (B8, B9+B10) 69.8% 66.0% 73.1% bifurcation (B8+B9, B10) 12.6% 13.6% 11.8% trifurcation (B8, B9, B10) 17.6% 20.4% 15.1%

From: Hoffman, Barr et al CT-Based Quantitative Lung Measures in a Healthy Multiethnic Sample: The MESA Lung Study (in review)

% Emphysema in Normals – Linear Model: From MESA Lung

Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Page 61: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

From: Hoffman, Barr et al CT-Based Quantitative Lung Measures in a Healthy Multiethnic Sample: The MESA Lung Study (in review)

% Emphysema in Normals – Linear Model: From MESA Lung

% Emphysema in Normals – Linear Model: From MESA Lung

From: Hoffman, Barr et al CT-Based Quantitative Lung Measures in a Healthy Multiethnic Sample: The MESA Lung Study (in review)

Translating XeCT to Human Imaging

Non-Smoker Smoker

Xe-CT Differentiates between Normal Smokers and Normal Non-Smokers

Percentage of lung that does not curve fit

  Non-Smoker (N=10) 23.78% w/o Fit

  Smoker (N=3) 40.71% w/o Fit

Time Constants (sec)

M Fuld, EA Hoffman et al. ATS. 2009 San Diego, Abst 3530

Assessment of Regional Heterogeneity of Parenchymal Perfusion

J Appl Physiol. 2006 Nov;101(5):1451-65

Saline Lavage: HPB-Based Blood Flow Redistribution

EA Hoffman et al. Proc Am Thorac Soc. 2006 Aug;3(6):519-32

Carol
Typewritten Text
Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Page 62: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

LPS Injury: Inhibition of HPB-Based Blood Flow Redistribution

EA Hoffman et al. Proc Am Thorac Soc. 2006 Aug;3(6):519-32

Normal vs. Pathologic Heterogeneity of Pulmonary Blood Flow

Proc Natl Acad Sci U S A. 2010 Apr 20;107(16):7485-90

NS SNI SCE

Mean Transit Time Deviations from Normal as a Phenotype of Emphysema Susceptibility

Proc Natl Acad Sci U S A. 2010 Apr 20;107(16):7485-90

Dual-Energy CT: Perfused Blood Volume (PBV) Material Decomposition to Identify Iodine Concentration

Fuld MK et al Radiology2013 Jun;267(3):747-56.

Dual Energy Imaging of Perfused Blood Volume

Perfused Blood Volume Pre vs. Post Sildenafil Response

Normal vs. Phenotype of Empysema Susceptible

NS SS

In collaboration with Krishna Iyer, John Newell Jr., and Sif Hansdottir

Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Carol
Typewritten Text
Page 63: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Regional Coefficients of Variation Pre and Post Sildenafil Normal vs. CE Susceptible Smoker

In collaboration with Krishna Iyer, John Newell Jr., and Sif Hansdottir

Total Pulmonary Vascular Volume

Assessment of Regional Lung Mechanics 1) Matching Small Scale Structures 2) Measuring Small Scale Changes

•  Two lung CT scans are done at two different times at full inspiration (TLC).

•  Local points are optimally matched by image registration to assess changes in small scale lung biomechanic signal.

Time 0 (flexible)

Time 1 (stationary)

+

+

Image Registration

Small Scale Changes

∆v +

∆v

Matched Small Scale (Acinar) Structures

In collaboration with Jiwoong Choi, John Newell Jr., and Ching-long Lin, M Milhem, T Knutson, J Tessier, M Kansy

Time 0 (flexible)

Time 1 (stationary)

+

+

Image Registration

Normal Control (Non-smoker) Very little change over time in the small scale biomechanic signal

Time 0 (flexible)

Time 1 (stationary)

+

+

Positive Control (Asymptomatic Smoker) Very little change over time in the small scale biomechanic signal

∆* ≥ 50%

Image Registration ∆* ≥ 50%

100%

20%

100%

20%

In collaboration with Jiwoong Choi, John Newell Jr., and Ching-long Lin, M Milhem, T Knutson, J Tessier, M Kansy

Cancer Subject (Sarcoma) 1) Initial Scan No Lung Metastases

2) Followup Scan with Lung Metastases 3) Small Scale Change in Biomechanic Effect Noted In Several Areas

Including the Growing Lung Nodules Between the Two Scans

Pre (flexible)

Met (stationary)

+

+

∆* ≥ 50% Image Registration

100%

20%

In collaboration with Jiwoong Choi, John Newell Jr., and Ching-long Lin, M Milhem, T Knutson, J Tessier, M Kansy

Cluster Analysis

Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Page 64: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

CT Phenotypes Must Provide Insights into the Etiology of the Disease, Not Just Road

Maps of Destruction And….

In order to determine abnormal, one must first

establish normal!

Carol
Typewritten Text
Establishing a "Normal" Baseline in QCT Eric A. Hoffman, PhD University of Iowa Carver College of Medicine
Carol
Typewritten Text
Carol
Typewritten Text
Carol
Typewritten Text
Page 65: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Oral Presentations

1 Size and Collapsibility of Emphysema Holes on CT of COPD: Evaluation with a New Method SEO JB, Oh SY, Lee M, Kim N, Lee SM Purpose: Although the extent of LAA at CT of COPD is known to correlate with loss of lung function, the size and collapsibility of emphysema holes were rarely investigated. The purpose of this study is to assess the size and collapsibility of emphysema clusters with a new method and to correlate it with pulmonary functional test (PFT). Methods and Materials: Volumetric, inspiration and expiratory CT scans in 60 COPD patients were selected. After extracting the area of LAA using the threshold of -950 HU, a new method of size classification of LAA, which is repeated application of multiple low pass filter with various size of Gaussian kernel, was used. Emphysema clusters were divided into four groups; E1<1.5mm, 1.5mm≤E2<7mm, 7mm≤E3<15mm, 15mm≤E4. The difference in distribution of each group between inspiration and expiration CT was assessed. The results were compared with PFT with Pearson correlation. Results: The mean extents of LAA, E1, E2, E3, and E4 were 27.2%, 3.5%, 12.9%, 6.8%, 3.9% on inspiration CT, 16.5%, 1.66%, 7.8%, 4.2%, 2.9% on expiratory CT, respectively. Among the four clusters, E3 showed larger decline on expiratory CT than other clusters. Both E2 and E3 were significantly correlated with FEV1(r=-0.449 and -0.423, respectively). Conclusion: Emphysema clusters of sublobular size were most collapsible. Emphysema of acinar and sublobular size contributes most to the FEV1 decline. This study shows that in addition to the extent of LAA, the size of emphysema clusters at CT of COPD should be also considered in relation to the loss of lung function.

2 Automated Nodule Segmentation With Sub-voxel Accuracy Using Mutual Interaction of Pulmonary Segmentation Structures OGUZ I, Raffy P, Wood S Objectives: Automated segmentation of nodules from CT scans is challenging due to border ambiguity with neighboring structures such as vessels and the pleura, as well as poor boundary contrast of more subtle nodules. We developed a novel method that robustly segments nodules by allowing mutual interaction between the nodule and surrounding structures, leading to a highly accurate segmentation. Methods/Materials: Our method relies on a graph-based segmentation algorithm for mutually interacting objects and provides sub-voxel accuracy by operating on the 3-D space rather than the image grid. It is robust to image noise and local minima. Our method was tested on solid nodules in the public Lung Imaging Database Consortium (LIDC) database and evaluated against the manual segmentation of 4 experts. The slice thickness ranges from 0.45 to 5mm (average voxel size: 0.70 x 0.70 x 1.69mm), with thin-slice scans defined as slice thickness ≤ 1.25mm. The tube current range is 40-582 mA and tube voltage range 120-40 kVp. Results: Average surface distance between the automated and manual segmentation was 0.71 ± 0.51mm (SD) (1,002 subjects, n = 1,307 nodules), with sub-voxel accuracy in 96% of cases. Performance was not significantly different between thin- and thick-slice scans (p>0.5): 0.72 ± 0.55mm (thick; n=494) vs. 0.70 ± 0.46mm (thin; n=508). Average surface distance was 0.65 ± 0.39mm for isolated nodules (n = 534), 0.80 ± 0.64mm for juxtapleural nodules (n = 446) and 0.69 ± 0.48mm for juxtavascular nodules (n = 327). Conclusions: Our novel nodule segmentation algorithm can robustly segment nodules with sub-voxel accuracy. Mutually interacting segmentation of the nodule and neighboring structures yields more accurate results and helps avoid “leaks” into the pleura or vessels.

Page 66: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Oral Presentations (Cont.)

3 Computer-aided Classification of Diffuse Lung Disease Opacities by Use of Sparse Representation-based Method on 3D-CT Images KIDO S, Zhao W, Xu R, Hirano Y Objectives: To improve a computer-aided diagnosis method for classification of diffuse lung disease (DLD) opacities such as consolidation, ground-glass opacity, honeycombing, emphysema, diffuse nodules and normal on 3D-CT images by use of sparse representation (SR) based-methods. Method/Materials: K-SVD is an algorithm of deciding dictionary of SR in a singular value decomposition approach, and orthogonal matching pursuit (OMP) is the canonical greedy algorithm for sparse approximation. A combination of K-SVD and OMP is used for typical SR-based method. It can achieve satisfied recognition rates. However, it needs a lot of calculation time (Tcal). So, we replaced the K-SVD by K-Means to train the dictionary, and simplified OMP by selecting the desired number of atoms at one time (OMP1). We proposed three SR-based methods: SR1 (K-SVD+OMP), SR2 (K-Means+OMP), SR3 (K-Means+OMP1). We compared proposed methods with published artificial neural network classifier based method (AN) and published nearest neighbor classifier based method (NN). 1161 volumes of interest were (VOIs) adopted as training data to optimize the parameters of methods and train each method, and 1046 VOIs were used as test data for evaluation. Results: The SR-based methods achieved satisfied results (SR1:96.1%, SR2:95.6%, SR3:96.4%), compared with published methods (AN:75.8%, NN:65.1%). Furthermore, when K-SVD was replaced by K-Means, Tcal of dictionary learning was reduced (SR1:13520s vs. SR2:241s, SR3:350s). And, when OMP1 was used as a substitute of OMP, Tcal of recognizing one VOI was also reduced (SR1:1.27s SR2:0.29s vs. SR3:0.13s). Conclusions: The SR-based methods can achieve high recognition rates for classification of DLD opacities, and Tcal of SRs can be reduced by adopting K-Means and OMP1.

4 Dose Modulation in an Anthropomorphic Chest Phantom and its Relative Effects when using Single and Dual Energy Scan Modes NEWELL JD, Sieren JP, Guo J, Levy J, Hoffman EA Objectives: Quantitative CT for the assessment of lung structure and function depends on accurate measures of lung CT densities. We have evaluated 4D dose modulation on a dual energy scanner in both the single (SE) and dual energy (DE) scan modes. Methods: We used an anthropomorphic phantom (Phantom Labs, Salem, NY) that included lung equivalent material and an air filled trachea region. A Siemens Definition FLASH was used in both SE and DE scan modes with and without CareDose. Analysis was via placement of 3D cylindrical ROI’s down the center of the trachea and within each lung. Statistical analysis included basic statistics, Welch’s t-test. Results: There was a significant difference in tracheal air mean attenuation (p < 0.0001) as well as in the right and left lung (RL and LL) material (p<0.0001) using SE compared to DE with/without CareDose. There were significant differences in right (p<0.05) and LL (p<0.01) material attenuation using SE compared to SE+CareDose. There were no (p>0.05) SE vs. SE+CareDose differences in tracheal air values. There is no significant difference in RL and LL material (p>0.05) DE compared to DE+CareDose. There were significant tracheal air (p <0.05) DE vs. DE+CareDose differences. Conclusion: We have previously demonstrated that in the DE mode, the scatter correction used in DE mode serves to bring mean tracheal air closer to the expected -1000HU value. In this study, use of SE+CareDose significantly altered attenuation values of the RL/LL material compared to SE. DE+CareDose significantly changed the attenuation values of tracheal air compared to DE. These data suggest that dose modulation must be used consistently throughout a study since it does change lung density values even in SE mode where scatter correction has been employed.

Page 67: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Oral Presentations (Cont.)

5 Lung Lesion Volume Measurements in Simulated Reduced-dose CT Scans YOUNG S, Kim G, McNitt-Gray M Objective: To illustrate how lung lesion volume measurements may be affected when using reduced-dose CT scans. Methods: For this IRB approved study, CT scan data from 37 patients were obtained and anonymized. We simulated reduced-dose scans by adding noise to the clinical raw CT data, obtained prior to reconstruction of the clinical CT scan. The simulation approach has two benefits: (1) it allows us to “scan” the same patient at any arbitrary dose level without any additional exposure, and (2) it produces reduced-dose scans with the identical nodule position as in the original scan. For each clinical scan, reduced-dose scans were generated at dose levels ranging from 25% down to 3% of our current clinical dose protocols. On the reconstructed image data sets, we used a semi-automated, in-house CAD program to contour the lung lesions at each dose level. To evaluate the resulting lesion volumes, we calculated agreement between clinical and reduced-dose volumes using a Bland-Altman-style approach to determine mean percent volume difference and 95% confidence intervals (CI ) on the differences. Results: Initial results showed 95% CI ranging from 30% at the 25% dose level to 38% at the 3% dose level. Increased variability at 3% dose may result from increased noise under ultra-low dose conditions. Conclusion: Lung lesion volume measurements appear to be fairly consistent down to extremely low dose levels. However, measuring lesion volume is only one part of the radiologist’s task. Our reduced-dose simulation software provides a foundation for further analyses of the impacts of dose level on reader performance. In the future, we will evaluate reader repeatability, which will help to more clearly separate the effects of reduced dose scans from reader variability.

6 Determination of the Optimal Time-Window to Perform Wide Volume Dual Energy CT in a Swine Model of Pulmonary Embolism (PE) JIMENEZ-JUAN L, Dey C, Moghe S, Mehrez H, Homampour S, Paul NS Objective: To determine the optimal time-window after contrast injection to generate Dual Energy (DE) Iodine Maps. Methods: A swine model of PE was used (3 subjects, 40 kg). Autologous thrombotic material was injected via an 11F right femoral sheath and the locations of the emboli were confirmed with CT pulmonary angiography. Using wide volume (320x0.5mm) CT (Aquilion ONE, TMS, Japan), 10 sequential DECT scans were acquired every 4s, after injecting 40mL of non-ionic intravenous contrast (Visipaque 320mg/mL, GE Healthcare) at 4 mL/s. Iodine maps were created using a commercially available software (Body Perfusion, TMS, Japan). 2 observers independently compared the iodine maps for all DE acquisitions to determine which DE pairs resulted in similar perfusion information. Qualitative assessment was performed using a 3 point score (1=excellent, 2=moderate, 3=poor match). The observers were blinded to the anatomical locations of the PE demonstrated on the CTPA study. Quantitative analysis of the iodine maps was assessed by cross-correlation in time of entropy measures of the maps. Results: Qualitatively, iodine maps at 13s and 17s demonstrated the best perfusion data. The mean visual score for the three subjects was 1.77. Quantitative analysis in two subjects demonstrated a strong correlation between the iodine maps at 13s and 17s, more so than iodine maps at 9s and 21s (p<0.05). In the third subject the optimal iodine maps were obtained at 17s. Conclusion: Using a porcine PE model and 40mL of CM injection at 4 mL/s, the optimal time window for DECT is ≤ 4s and occurs 13-17s post injection of contrast.

Page 68: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Quantitative CT Imaging of the Lung Scientific Posters

Saturday, March 15, 2014

Image Quality Comparison of Low KV Imaging in CT Pulmonary Angiography Bijan B, Belashabadi S, Davoodi M, Kabir A Aim: Comparing quality of images utilizing 80 KV versus 100 KV. Methods: In a non-randomized triple blind parallel quasi-experimental study, we evaluated 140 patients with suspicion for pulmonary embolism by 80 or 100 KV CT angiography. Image quality was judged by discerning opacified segmental and subsegmental pulmonary arteries. Results: Quality of images was better using 80 KV (OR, 2.08). Vascular enhancement was significantly higher in all main, segmental and subsegmental arteries (P< 0.001) in 80 KV group. Mean number of measurable segmental arteries was also significantly higher among 80 KV group. Conclusion: Despite increased image noise, visualization of segmental & sub-segmental pulmonary arterial branches was improved in cases with 80 KV in comparison to 100 KV.

Dynamic Lung Perfusion with Wide Volume CT: Effect of Arterial Input on Arterial Flow Maps Jimenez-Juan L, Dey C, Homampour S, Mehrez H, Paul N Objective: To determine whether the location of the arterial input within the pulmonary trunk, left or right pulmonary artery influences the accuracy of dynamic CT lung perfusion maps. Methods: 54 dynamic wide volume CT perfusion scans were performed in 3 pigs using 30s continuous data acquisitions, 100kV, 50mAs, and 320x0.5mm detector providing 160mm coverage. Each subject had 9 pre-pulmonary embolus (PE) and 9 post-PE dynamic perfusion scans in combination of 3 flow rates (4, 6, and 8mL/s) and 3 contrast volumes (0.4, 0.6, and 0.8mL/kg). Perfusion analyses were carried out using maximum slope. Tissue region of interest (ROI) was prescribed in the lung parenchyma, and arterial input ROIs were placed in the pulmonary trunk (PT), left main (LM), and right main (RM) pulmonary arteries. For each arterial input, Time Density Curves (TDCs) were determined and perfusion maps were generated. Qualitative assessment was performed by two thoracic radiologists (R1, R2). The LM and RM perfusion maps were compared to the PT, using a 3 point scale (1= mismatch, 2=matching perfusion defects but visual difference in color maps; 3=identical). Pearson’s rank correlatio n coefficient (r) was used for statistical comparison of the perfusion maps and TDC. Results: Quantitative analysis demonstrated a strong correlation between the TDCs (r>0.83, p<0.005) and the perfusion maps (r> 0.99, p<0.005) for all injection rates/volumes, regardless the arterial input location. Qualitative analysis ascertained good agreement between the perfusion maps and the visual scores; R1= 2.8 ± 0.4 (range 2-3), R2= 2.6± 0.5 (range 2-3). Conclusion: Accurate lung arterial flow maps can be generated by selecting the pulmonary trunk, the left or the right pulmonary artery as arterial input locations.

Page 69: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Posters (Cont.)

Dynamic Lung Perfusion with Wide Volume CT: Impact of Bolus Length on Arterial Flow Values Jimenez-Juan L, Dey C, Homampour S, Mehrez H, Paul NS Objective: To assess whether the bolus length impacts the evaluation of dynamic CT lung perfusion maps determined with single input maximum slope (SIMS) analysis in a pig model. Method: 54 dynamic wide volume CT perfusion scans were performed in 3 pigs (40kg) using 30s continuous data acquisitions, 100kV, 50mAs, and 320x0.5mm detector configuration. Each subject had 9 pre-pulmonary emboli (PE) and 9 post-PE dynamic perfusion scans using 3 contrast volumes (16, 24, and 32mL) and 3 flow rates (4, 6, and 8mL/s). Perfusion analyses were performed using the SIMS approach and arterial flow (AF) perfusion maps were generated. For each contrast volume, the perfusion maps generated at 4 and 6 mL/s were compared to those acquired at 8 mL/s. Mean AF values and perfusion defects (PD) were determined. Spearman rank correlation of mean AF was calculated. Sensitivity and specificity to detect PD were also calculated. Results: Mean±SD AF values (mn-1) for 4, 6, and 8mL/s were 2.35±0.65, 2.73±0.76 and 2.91±0.65, respectively. There was strong correlation of mean AF values using 32mL of contrast volume at 6 and 8 mL/s, r=0.89, p=0.03. The sensitivity to detect PD was 0.78± 0.07 (0.67-0.84) using a contrast volume of 32mL at 6 mL/s and 0.64±0.14 (0.42-0.82) using a contrast volume of 32mL at 8 mL/s. The specificity to detect PD using a contrast volume of 32 mL at both 6 mL/s and 8 mL/s was>0.82. However, the sensitivity to detect PD using contrast volumes of 16 and 24 mL at any flow rate was <0.64. Conclusion: Optimal AF perfusion maps require an injection rate ≥6ml/s.

Validation of Simulated Tomosynthesis: Can it Replace Real Imaging for First Line Clinical Trials? Kicska G Objective: Digital tomosynthesis (DTS) is a relatively new imaging modality with multiple potential, yet to be proven, clinical applications. Unfortunately, clinical studies that test these applications are time-consuming, expensive and involve radiating volunteers. We report and validate a technique that produces simulated DTS (sim-DTS) images that may substitute for real DTS images in early clinical studies, thus accelerating identification of clinical applications. Materials and Methods: CT voxel phantoms were created using phantom and human CT data. These voxel phantoms were used to create simulate DTS images. The same phantoms and humans were imaged with real tomosynthesis. Real-DTS and sim-DTS images were then compared for out-of-plane blurring, signal to noise, modulation transfer function and image quality. Results: Out-of-plane blurring was near identical between sim and real DTS studies. Sim-DTS images have greater signal to noise compared to real-DTS (82.1 vs. 5.2). Modulation transfer function (MTF) is decreased for sim vs real-DTS images (36% vs 79% @ 4 lp/mm). Qualitatively, images from real and simulated studies are similar. Conclusions: Sim-DTS images produce near identical out-of-plane blurring compared to a real-DTS images, a major imaging feature of tomosynthesis. Sim-DTS images produced lower MTF response, a result of CT voxel phantom resolution. However, this is partly offset by higher signal to noise of sim-DTS images. These findings suggest that simulated exams can replace real-DTS images in certain first-line clinical investigations, thus identifying useful clinical applications faster, cheaper and without radiating patients. Applications deemed promising can be further investigated with a conventional clinical trial using only real-DTS.

Page 70: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Posters (Cont.)

Feasibility of Cardiac Output Measurements Based on CT Timing Bolus Imaging Using a Modified Timing Bolus Scalzetti EM, Rajebi H, Ogden KM Objectives: One way of selecting the optimal delay between contrast injection and CT pulmonary angiography (CTPA) is to obtain timing bolus images. The timing bolus method has been shown to allow estimation of cardiac output (CO), based on attenuation measurements in the aorta or main pulmonary artery (MPA) and application of the indicator dilution principle, fitting a gamma-variate curve to the time-attenuation data. We assessed the feasibility of calculating CO after we modified the volume of the timing bolus. Methods: Patients referred for evaluation of pulmonary embolism were identified retrospectively, with permission of the local IRB. In each case, 75 ml of a saline-contrast mixture were administered intravenously at a rate of 5 ml/s as a modified timing bolus, using a dual-headed power injector. Beginning at 4 s, 40 mm CT volumes were obtained every 2 s until contrast washed out of the MPA. Attenuation measurements were made in the MPA and aorta. This was followed by the CTPA, for which 75 ml of contrast were given at a rate of 5 ml/s. Separately, the relationship between CT attenuation and iodine concentration was assessed in a phantom. The time-attenuation data was modeled with a gamma-variate curve, and the CO calculated. Results: CT attenuation varied linearly with iodine concentration (slope 26 HU/mg/ml). The average r2 goodness of fit for the gamma-variate model was 0.90 (95% CI 0.74-0.97). The average CO was 4.3 L/min (95% CI 2.9-8.5), compared to a reported normal range of 4.0-8.0 L/min. Conclusions: Reasonable values for CO were obtained by applying an indicator-dilution approach to a modified timing bolus of contrast, the volume of which is equal to the contrast volume given for the CTPA. Continued work on validation seems to be warranted.

Reproducibility of Cardiac Output Measurements Based on CT Timing Bolus Imaging Using a Modified Timing Bolus Scalzetti EM, Rajebi H, Ogden KM Objectives: Cardiac output (CO) can be calculated from measurements made in the main pulmonary artery (MPA) on timing bolus images that are obtained as part of a CT pulmonary angiogram (CTPA). We implemented this indicator dilution-based method with a modified timing bolus. We identified patients who had two or more CTPAs but lacked acute abnormalities, and compared the calculated COs. Methods: Patients referred for pulmonary embolism evaluation were identified retrospectively, with permission of the local IRB. In each case, 75 ml of a saline-contrast mixture were administered intravenously at a rate of 5 ml/s as a modified timing bolus, using a dual-headed power injector. Beginning at 4 s, 40 mm CT volumes were obtained every 2 s until contrast washed out of the MPA. Attenuation measurements were made in the MPA and aorta. This was followed by CTPA, for which 75 ml of contrast were given at a rate of 5 ml/s. The attenuation of unenhanced blood in the aorta was subtracted from the MPA attenuation values. The time-attenuation data was modeled with a gamma-variate curve, and the CO calculated. Paired values of CO were compared (SigmaPlot). Results: 4 patients had 2 CTPAs and 1 patient had 3 CTPAs during the study period, yielding 6 CO pairs. All pairs but one differed by <10%. Averages of paired values did not correlate with differences within pairs (slope of regression line 0.005). The average r2 goodness of fit for the gamma-variate model was 0.85 (range 0.72-0.96). Conclusions: The method, as applied to the modified timing bolus, had good reproducibility in this small sample of patients who had repeat examinations made under similar clinical conditions. Differences do not appear to track with the magnitude of CO. Continued work on validation seems to be warranted.

Page 71: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Posters (Cont.)

Distensibility Differs Regionally in Asthma vs. Normal Controls: A CT Investigation in SARP subjects Shim SS, Schiebler ML, Sorkness RL, Jarjour NN, Kanne J, Fain SB Purpose: To assess the change of airway lumen area from the 1st-6th generations on inspiratory and expiratory computed tomography (CT) in asthmatics as compared with healthy normal subjects. Methods: We enrolled 185 SARP patients (152 asthmatics and 33 controls). There were corresponding to 3681 airway segments in asthmatics and 820 in normal subjects evaluated. The difference between lumen areas was calculated for the 1-6th generation of the airway tree in CT images using VIDA software, acquired at total lung capacity (LAt) and functional residual capacity (LAf). Distensibility was defined by following equation: 100 x [1-(LAt/LAf)]. Differences in distensibility in 1-6th generations of the airway tree was then compared for asthma vs. control groups. Results: Mean value of distensibility of all generations differed and was 20.6% in asthmatics and 16.9% in normal subjects (p<0.001). By generation, distensibility of asthmatics for 1st-6th generation respectively was: 15.7%, 16.2%, 20.4%, 22.8%, 20.3% and 18.7%, whereas for normal controls was: 13.3%, 16.7 %, 15.6%, 16.2% 17.8%, and 20.6 %. Differences at the 3rd (p=.048), 4th (p<.001), and 5th (p=.124) generations were significant or trending towards significance. Qualitatively, a steady increase of distensibility was observed from central to more distal airway generations for normal controls, whereas, distensibility in asthmatics this steady increasing trend with generation only held for the 1-4th generations decreasing thereafter in the more distal 5-6th generations. Conclusion: Asthmatics demonstrate a larger distensibility than normal subjects, particularly in 3rd and 4th airway generations and demonstrate a pattern of decreasing airway lumen change in the more distal airways when compared to normal subjects.

Airway Distensibility in Asthma Patients by Computed Tomography: Correlation with Clinical Indices and Reversibility after Bronchodilator Therapy Shim SS, Schiebler ML, Sorkness RL, Jarjour NN, Denlinger LC, Fain SB Purpose: The aim of this study is to assess airway distensibility found on computed tomography (CT) in asthma patients and to explore correlation with clinical indices and response to bronchodilator therapy. Methods: We prospectively enrolled a total of 152 Severe Asthma Research Protocol (SARP) subjects. The airway distensibility was defined as the ratio of difference between airway lumen area found at CT during total lung capacity (TLC) and functional residual capacity/ lumen area at TLC. Distensibility of all 1st-6th generation airways was measured using VIDA software, and the mean values for each subject were calculated. Correlations were made between mean value of distensibility and clinical indices including sex, age, body mass index (BMI), severity, and FEV1 pre- and post-bronchodilator therapy to determine reversibility. Results: Univariate analysis showed significant correlation with airway distensibility for the following clinical indices: BMI (r=0.271, p=0.001), pre-bronchodilator FEV1 %predicted (r=0.160, p=0.048), pre-bronchodilator FEV1/FVC (r=0.209, p=0.01), post-bronchodilator FEV/FVC (r=0.201, p=0.015) and max bronchodilator FEV1 reversibility (r= - 0.168, p=0.039). Using a multivariate regression analysis, average airway distensibility in asthmatics was dominated by BMI (p<0.001). BMI in females was also found to be highly correlated with distensibility, but this was not found for male subjects. Conclusion: In SARP patients, increased airway distensibility is positively associated elevated BMI in females and negatively associated with the % reversibility of bronchodilator therapy. In combination, these findings suggest increased distensibility may help distinguish different patterns of air flow obstruction in asthma.

Page 72: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

Scientific Posters (Cont.)

Lobar Contribution to Lung Volume and Heterogeneity in Healthy Subjects: Quantitative Assessment by CT Silva M, Nemec S, Dufresne V, Bankier AA Objectives: To quantify the contributions of individual lobes to lung volume, and to characterize quantitative morphological metrics associated with these volumes. Methods/Materials: 17 healthy individuals underwent spirometrically monitored volumetric chest CT at total lung capacity (TLC), functional residual capacity (FRC), and mean inspiratory capacity (MIC), measured half-way between TLC and FRC. A semiautomatic software for parenchymal quantification (Pulmo3D, MeVis, Germany) was used to measure lobar volume, lobar density (MLoD and MLoD%) and the standard deviation of density, as a surrogate parameter for parenchymal heterogeneity. Differences in lobar metrics between individual lobes, and between lung volumes were assessed with analyses of variance for repeated measurements and with unpaired T tests. Results: Lobar contributions to TLC were: RUL 17%; RML 8%; LUL 21%; RLL 27%; LLL 27%. Lobar contributions to MIC were: RUL 18%; RML 9%; LUL 22%; RLL 26%; LLL 26%. Lobar contributions to FRC were: RUL 18%; RML 11%; LUL 25%; RLL 24%; LLL 23%. At TLC, the relative contribution of the lower lobes to lung volume was bigger than at FRC (p<0.001), while reverse phenomenon was observed for the upper lobes (p<0.001). The lower lobes were significantly denser than all other lobes, notably at FRC (ΔMLoD 76 ± 55 HU, p<0.001). Heterogeneity was higher in the lower lobes than in all other lobes, again, notably at FRC (p<0.001). Conclusions: The lower lobes make the biggest contribution to overall lung volume and, at any lung volume, are denser and more heterogeneous than other lobes. Our results confirm the feasibility of in-vivo assessment of lobar metrics and provide normative data of lobar volumes, density, and heterogeneity in healthy volunteers.

Page 73: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

2014 Quantitative CT Imaging of the Lung Disclosures As a provider accredited by ACCME, the Society of Thoracic Radiology must ensure balance, independence, objectivity and scientific rigor in its educational activities. Course Director(s), Planning Committee Members, Faculty, and all others who are in a position to control the content of this educational activity are required to disclose all relevant financial relationships with any commercial interest related to the subject matter of the educational activity. Safeguards against commercial bias have been put in place. Faculty will also disclose any off-label and/or investigational use of pharmaceuticals or instruments discussed in their presentation. Disclosure of these relevant financial relationships will be published in course materials so those participants in the activity may formulate their own judgments regarding the presentation. Listed below are the Quantitative CT Imaging of the Lung Course Course Directors, Review Committee Members, Faculty and Scientific Oral Presenters and their co-authors:

1) who have disclosed and the potential conflict has been resolved.

Full Name Commercial Interest What was Received Role Sean B. Fain, PhD GE Healthcare Grant Support Principal

Investigator Xerned LLC Honorarium Scientific Advisory

Board GE Healthcare Research Grant Principal

Investigator Xemed LLC Honorarium Scientific Advisor

Jonathan Goldin, MBChB, PhD MedQia Ownership Board Member

Junfeng Guo, PhD Share holder in VIDA diagnostics, a

company commercializing lung image analysis software developed, in part, at the University of Iowa.

nothing received within the past 12 months

share holder

Eric A. Hoffman, PhD VIDA Diagnostics Stock My lab utilized data analysis software from VIDA, a company for which I am a founder and share holder

Jeffrey Kanne, MD Perceptive Informatics Consulting Fee Independent Consultant (not relevant to material at hand)

Namkug Kim, PhD None None Software developing and experimental study

Joshua Levy, BS Image Owl, Inc. The Phantom Laboratory is a stock holder

Manager

The Phantom Laboratory, Incorporated

I am an employee and stock holder

President and Responsible for Engineering and Research and Development

Page 74: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

David A. Lynch, MB Perceptive Imaging, Boehringer Ingelheim, Genentech, Veracyte, Gilead, Intermune

Consulting Fee Consultant

Centocor, Siemens, NHLBI Research Support Researcher

Michael McNitt-Gray, PhD Consultant, Fulbright and Jaworski, LLC

Consultant Fees Consultant

Flaherty Sensabaugh Bonasso PLLC Consultant Fees Consultant Siemens Healthcare Institutional Research

Support

Investigator

Siemens Healthcare Grant Support Investigator Hatem Mehrez, PhD Toshiba Medical Systems, Canada Salary Clinical Research

Physicist

Sachin Moghe, PhD Toshiba Medical Research Institute, USA

Salary Research

John D. Newell, Jr., MD VIDA Diagnostics Money, Stock Options Consultant, Advisory Board

Siemens Health Research Grant Research Support Siemens HealthCare AG Research Money. Travel

Money Researcher

VIDA Diagnostics Inc. Money for Consulting. Stock Options.

Paid Consultant

Ipek Oguz, PhD VIDA Diagnostics Salary Employee

Yoshiharu Ohno, MD, PhD Toshiba Medical Sustems Corporation

Research Grant Principal Investigator

Philips Electronics Japan Research Grant Principal Investigator

Bayer Pharma Research Grant Principal Investigator

Guerbet Japan Research Grant Principal Investigator

Narinder Paul, MD Research Support Toshiba Medical Systems

Reserach grant PI

Philippe Raffy, PhD I am an employee of VIDA Diagnostics Inc. Co-author

Joyce D. Schroeder, MD Siemens Corp. Research Salary Support Principal Investigator on Siemens-funded CT Study

Joon Beom Seo, MD Guerbet Korea Honorarium Speaker Siemens Honorarium Speaker

Susan Wood, PhD VIDA Diagnostics Employee CEO

Page 75: Quantitative CT Imaging of the Lung Grand Hyatt San ... QCT... · Dose Modulation in an Anthropomorphic Chest Phantom and its Relative ... Young S, Kim G, ... 1995, Coxson AJRCCM

2) who declared no relevant financial relationship(s) with industry

Soudabeh Belashabadi, MD, MBA Bijan Bijan, MD, MBA Mohammad Davoodi, MD Pim de Jong, MD, PhD Loren Denlinger, MD, PhD Chris Dey, MD Valarie Dufresne, MD Raul San Jose Estepar, PhD Yasushi Hirano, PhD Shabnam Homampour, MASc Nizar Jarjour, MD Laura Jimenez-Juan, MD Philip F. Judy, PhD Ali Kabir, MD, PhD Shoji Kido, MD, PhD Grace Hyun Kim, PhD Sang Min Lee, MD Minho Lee, MS Francesco Molinari, MD Stefan Nemec, MD Kent Ogden, MD Sang Young Oh, MD Hamid Rajebi, MD Geoffrey D. Rubin, MD Ernest Scalzetti, MD Mark Schiebler, MD Sung Shine Shim, MD Jered Sieren, BS Mario Silva, MD Ronald Sorkness, PhD Daniel C. Sullivan, MD George R. Washko, MD Rui Xu, PhD Stefano Young, PhD Wei Zhao, MS