QF1-012 - Multi-centre evaluation of diffusion tensor imaging (DTI) in an isotropic phantom: Towards validation of cardiac DTI sequences.

Thursday, February 13, 2020

3:18 PM – 3:21 PM

Location: Sabal/Sago (Meeting Room in Exhibit Hall)

Background: Cardiac diffusion tensor imaging (DTI) continues to develop as a non-invasive contrast-free method for in vivocharacterization of cardiac microstructure. Currently, data from healthy subjects^1,2,3,4report a wide range for both mean diffusivity (MD, 0.75 × 10^-3to 1.72 × 10^-3mm²/s) and fractional anisotropy (FA, 0.29 to 0.61). Consequently, there is a need to better understand the variation in DTI results reported in the literature that may arise from differences in scanners, field strengths, pulse sequences and post-processing. The objective of this multi-centre study was to investigate the inter-site variation in DTI parameters in an isotropic phantom, as a first step towards harmonisation of cardiac DTI methods.

Methods: 11 sites participated in the study: Philips (N=1) and Siemens (N=11); 1.5T (N=1), 3T (N=10) and 7T (N=1). Phantoms comprising 7 tubes filled with 0 - 20% polyvinylpyrrolidone (PVP)⁵were prepared at the University of Leeds and shipped to participating sites (Fig 1). The phantoms were chilled in ice water and imaged at 0°C. DTI scans were performed using product spin echo EPI with parameters matching typical cardiac DTI protocols: TR / TE = 3000 / 85 ms, in-plane resolution = 2.5mm, slices = 3, thickness = gap = 8mm, b_low= 0 s/mm², b_high= 450 s/mm², number of diffusion directions = 6, bandwidth ~3000Hz, parallel imaging = 2x, and triggered with simulated ECG = 60 bpm. Data storage and management were performed using the Human Heart Project, an online platform for heart imaging research^6,7. CMRDiffTools was used for image processing. To obtain a reference diffusivity (D_ref), the self-diffusion of water as measured using an independent non-MR technique, was extrapolated to 0°C using a 2^ndorder polynomial fit⁸.

Results: Representative MD and FA maps (Fig 1), and MD and FA across tubes and sites are shown (Figs 2 and 3). At 0% PVP, the average MD (mean ± SD across sites) was 1.148 × 10^-3± 0.030 × 10^-3mm²/s and its coefficient of variation was 1.8 ± 1.3 %; the average FA was 0.033 ± 0.012 and its coefficient of variation was 46.4 ± 15.3 %. D_ref(H₂O) extrapolated from Mills⁸ was 1.113 × 10^-3mm²/s.

Conclusion: A prerequisite for the establishment of cardiac DTI-based biomarkers is the assessment of the sources of variation in DTI data. We present initial results, acquired at 11 sites with a standardised phantom, acquisition, and post-processing pipeline, that show the variation due to differences in scanners and sites. The average MD (0% PVP) was in good agreement with D_ref(H₂O). The inter-site variation falls below the variation of MD and FA reported in vivo, pointing to the additional impact of other variables in vivo, such as pulse sequences and diffusion times^3,9. One consideration is that the product sequences used are not motion-compensated and therefore unsuitable for in vivo studies¹⁰. Future work will investigate the additional variation arising from the use of site-specific pulse sequences designed for in vivo cardiac DTI.

Presenting Author(s)

IT

Irvin Teh, PhD

Senior Research Fellow
University of Leeds

Disclosure: Disclosure information not submitted.

Primary Author(s)

IT

Irvin Teh, PhD

Senior Research Fellow
University of Leeds

Disclosure: Disclosure information not submitted.

Co-Author(s)

WR

William A. Romero, MSc

University of Lyon

Disclosure: Disclosure information not submitted.
Erica Dall'Armellina, MD, PhD

Consultant Cardiologist
Department of Biomedical Imaging sciences, University of Leeds

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.
Daniel B. Ennis, PhD

Associate Professor
Stanford University

Disclosure: Disclosure information not submitted.
Pedro F. Ferreira, PhD

Physicist
The Royal Brompton Hospital

Disclosure: Disclosure information not submitted.
PK

Prateek Kalra

Research Associate Engineer
Ohio State University wexner medical center

Disclosure: Disclosure information not submitted.
AK

Arunark Kolipaka, PhD

Associate Professor and Director of MRI
The Ohio State University

Disclosure: Disclosure information not submitted.
Sebastian Kozerke, PhD

Professor of Biomedical Imaging
University and ETH Zurich

Disclosure: Disclosure information not submitted.
DL

David Lohr, MSc

Comprehensive Heart Failure Center

Disclosure: Disclosure information not submitted.
KM

Kevin Moulin, PhD

Post-doctoral Fellow
Stanford University

Disclosure: Disclosure information not submitted.
Christopher Nguyen, PhD

Faculty
Massachusetts General Hospital and Harvard Medical School

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.
Sonia Nielles-Vallespin, PhD

Senior Lecturer
Imperial College of London

Disclosure: Disclosure information not submitted.
BP

Beau Pontre, PhD, BSc

The Univeristy of Auckland

Disclosure: Disclosure information not submitted.
LS

Laura M. M. Schreiber, MBA, PhD, MBA

Comprehensive Heart Failure Center

Disclosure: Disclosure information not submitted.
Andrew D. Scott, PhD

Physicist
The Royal Brompton Hospital

Disclosure: Disclosure information not submitted.
DS

David Sosnovik, MD

Massachusetts General Hospital, Harvard Medical School

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.
Christian T. Stoeck, PhD

Junior Group Leader
University and ETH Zurich

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.
CT

Cyril Tous

Université de Montréal

Disclosure: Disclosure information not submitted.
ET

Elizabeth M. Tunnicliffe, PhD

Postdoctoral researcher
University of Oxford

Disclosure: Disclosure information not submitted.
VW

Vicky Wang, PhD

Postdoc
San Francisco Veteran Affairs Medical Center

Disclosure: Disclosure information not submitted.
AW

Andreas M. Weng

University Hospital Wuerzburg

Disclosure: Disclosure information not submitted.
Alistair Young, PhD

Professor
King's College London

Disclosure: Disclosure information not submitted.
Pierre Croisille, MD, PhD

Professor
Jean-Monnet University

Disclosure: Disclosure information not submitted.
Magalie Viallon, PhD

Researcher
Univ Lyon, UJM-Saint-Etienne, INSA, CNRS UMR 5520, INSERM U1206, CREATIS

Disclosure: Disclosure information not submitted.
Jurgen E. Schneider, PhD

Professor of Biomedical Imaging
University of Leeds

Disclosure: Disclosure information not submitted.