Hannes Dillinger¹, Charles McGrath¹, and Sebastian Kozerke^1
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland

This work employs spectral interpretation of velocity encoding gradients (VEG) to assess their encoding power of fluctuating velocities in turbulent flows. Using CFD and MRI simulations based on particle tracking, it is shown that Reynolds Stress Tensor values are underestimated for VEG frequencies typically used in 4D Flow MRI, while mean velocity estimation is not impacted.

Carola Fischer¹, Jens Wetzl¹, Tobias Schäffter^2,3,4, and Daniel Giese^1
1Magnetic Resonance, Siemens Healthcare GmbH, Erlangen, Germany, ²Physikalisch-Technische-Bundesanstalt (PTB), Braunschweig and Berlin, Germany, ³Department of Medical Imaging, Technical University of Berlin, Berlin, Germany, ⁴School of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom

Background phase correction is necessary to correctly quantify flow velocity values from 2D phase-contrast MR images. However, typically used static tissue correction is susceptible to wrap-around resulting in even larger quantification errors. In this work, we successfully implemented a robust and automatic background phase correction algorithm based on M-estimate Sample Consensus (MSAC). MSAC achieves robust results over wide ranges of its few parameters. Based on 49 phase-contrast time series with and without wrap-around, MSAC reduced the average root-mean-squared error from 1.71±0.34cm/s (static fit correction) to 0.78±0.07cm/s (MSAC) in presence of wrap-around. 

Jihye Jang^1,2, Yansong Zhao¹, Jouke Smink³, Andrew J Powell², and Mehdi H Moghari^2
1Philips Healthcare, Gainesville, FL, United States, ²Department of Pediatrics, Harvard Medical School, Boston, MA, United States, ³Philips Healthcare, Best, Netherlands

Phase-contrast (PC) cine MRI is used for the clinical assessment of blood flow in various cardiovascular diseases. One of the challenges of PC is a trade-off between velocity aliasing artifacts and velocity-to-noise ratio (VNR). To improve VNR without velocity aliasing, we implemented a novel dual-venc dual-echo 2D cine PC sequence where both high and low-venc data are acquired within a single TR and velocity is measured using data from both venc images. In 10 patients, the dual-venc PC sequence demonstrated similar blood flow measurements with significantly better VNR compared to the standard single-venc PC sequence.

Ali Nahardani^1,2, Simon Leistikow^2,3, Martin Krämer^1,4, Karl-Heinz Herrmann¹, Wan-Ting Zhao^1,2, Daniel Güllmar¹, Lars Linsen³, Jürgen R. Reichenbach¹, and Verena Hoerr^1,2,5
1Medical Physics Group, Institute of Diagnostic and Interventional Radiology, Jena University Hospital - Friedrich Schiller University Jena, Jena, Germany, ²Center for Sepsis Control and Care, Jena University Hospital, Jena, Germany, ³Institute of Computer Science, Department of Mathematics and Computer Science, Westfälische Wilhelms-Universität Münster, Muenster, Germany, ⁴Institute of Diagnostic and Interventional Radiology, Jena University Hospital - Friedrich Schiller University Jena, Jena, Germany, ⁵Clinic for Radiology, University Hospital Muenster, Muenster, Germany

Compressed Sensing (CS) is a popular reconstruction technique supporting 4D-flow MRI acquisitions. Many literature studied the velocity changes after CS simplifying the velocity vector field into a scalar field along the time domain. The aim of our investigation was to assess how CS influences reconstructed velocity vector fields in space. Our results showed that CS underestimated the maximum velocity values, broadened the full-width-at-half-maximum of the velocity profiles, and preserved the directional information of the velocity vector fields compared to L2-ESPIRiT. The results of CS were in agreement for differently undersampled data, while the L2-ESPIRiT reconstruction provided differing outputs.

Carmen P S Blanken¹, Lukas M Gottwald², Jos J M Westenberg³, Eva S Peper², Bram F Coolen⁴, Gustav J Strijkers⁴, Aart J Nederveen², R Nils Planken², and Pim van Ooij^2
1Radiology and Nuclear Medicine, Amsterdam UMC, Amsterdam, Netherlands, ²Radiology and Nuclear Medicine, Amsterdam UMC, location AMC, Amsterdam, Netherlands, ³Radiology, Leiden UMC, Leiden, Netherlands, ⁴Biomedical Engineering and Physics, Amsterdam UMC, location AMC, Amsterdam, Netherlands

We compare pseudo-spiral undersampled whole-heart 4D flow MRI with CS reconstruction with a clinically used EPI readout in a cohort of healthy subjects and patients with valve regurgitation. Our results indicate that pseudo-spiral CS 4D flow MRI is at least as reliable as EPI-based 4D flow MRI in terms of inter-valve consistency of blood flow measurements and agreement with 2D MRI-based regurgitant volume measurement. Doubling the undersampling factor of the CS acquisition results in <10% deviation of the measurements compared to the original acquisition, suggesting that CS scan times may be shortened to expedite clinical implementation.

Jos J.M. Westenberg¹, Hans C van Assen¹, Pieter J van den Boogaard¹, Jelle J Goeman¹, Hicham Saaid², Jason Voorneveld³, Johan Bosch³, Sasa Kenjeres⁴, Tom Claessens², Pankaj Garg⁵, Marc Kouwenhoven⁶, and Hildo J Lamb^1
1Leiden University Medical Center, Leiden, Netherlands, ²Ghent University, Ghent, Belgium, ³Erasmus Medical Center, Rotterdam, Netherlands, ⁴University of Technology Delft, Delft, Netherlands, ⁵Norwich University Hospital, Norwich, United Kingdom, ⁶Philips Healthcare, Best, Netherlands

Echo Planar Imaging (EPI) is associated with inaccurate velocity quantitation in 4D flow MRI. The systematic errors depend on the orientation of readout and blip phase encoding gradient with respect to the flow direction. This study evaluates EPI-related errors in flow rate and velocity quantitation for in vivo intracardiac 4D flow MRI in a phantom and in ten healthy volunteers. Errors in median flow rate and velocity remain below 10% for normal in vivo intracardiac flow with EPI factor 5. The error is smallest when readout and blip phase encoding gradients are both perpendicular to the main flow.

Justin Baraboo¹, Michael Scott¹, Haben Berhane¹, Ashitha Pathrose¹, Michael Markl¹, Ning Jin², and Kelvin Chow^1,2
1Northwestern, Chicago, IL, United States, ²Cardiovascular MR R&D, Siemens, Chicago, IL, United States

4D-Flow MRI is a valuable technique for quantifying cardiovascular hemodynamics in the aorta; however, it suffers from manual off-line post processing. To address this, we integrated our custom deep learning tools for automatic 4D-Flow processing within the on-scanner reconstruction environment through Siemen’s Framework for Image Reconstruction (FIRE) interface. We retrospectively reconstructed raw data from 10 patients with aortic dilation, valve repair and/or aneurysm as well as one, prospectively recruited, control on scanner. Our deep learning tools ran successfully, and an aortic velocity maximum intensity projection cine was generated and sent to the scanner’s console alongside the reconstructed 4D-flow.

Jeesoo Lee¹, Liliana Ma¹, Michael Baran Scott¹, Alexander Jonathan Barker², James David Thomas³, and Michael Markl^1
1Radiology, Northwestern University, Chicago, IL, United States, ²Radiology, University of Colorado Denver Anschutz Medical Campus, Aurora, CO, United States, ³Cardiology, Northwestern University, Chicago, IL, United States

4D flow MRI can measure the full 3D mitral regurgitant jet velocity field allowing for direct characterization of jet flow dynamics. We performed in vitro investigation of two fluid dynamic phenomena known as flow entrainment and axial momentum conservation for MR-mimicking pulsatile flow jet using 4D flow MRI. The impact of spatial resolution on the characterization accuracy was also systematically assessed. The results revealed that 1) jet flow volume may not be equivalent to regurgitant flow volume and 2) axial momentum could reliably characterize MR jet by avoiding partial volume effect close to the orifice.

Eric Schrauben¹, Mitzi van Andel², Lukas Gottwald¹, Aart Nederveen¹, Maarten Groenink^1,2, and Pim van Ooij^1
1Department of Radiology & Nuclear Medicine, Amsterdam University Medical Centers, location AMC, Amsterdam, Netherlands, ²Department of Cardiology, Amsterdam University Medical Centers, location AMC, Amsterdam, Netherlands

Aortic pulse wave velocity (PWV), which can be an indicator of certain cardiovascular disease via aortic stiffness estimation, requires high temporal resolution for accurate measurements. Using compressed sensing based 4D flow MRI of the aorta,  we propose a tool for assessment of PWV. The impact of data requirements (and thereby scan time) were investigated via retrospective undersampling in a cohort of healthy volunteers. The findings were then applied to Marfan subjects to determine if the minimal data requirements effectively detect differences between patients and healthy controls.

Tilman Stephan Emrich^1,2,3, Natalie Ring², U. Joseph Schoepf¹, Ning Jin⁴, Daniel Sebastian Dohle⁵, Fei Xiong⁴, Anna Lena Emrich^5,6, Karl-Friedrich Kreitner², and Akos Varga-Szemes^1
1Department of Radiology and Radiological Science, Medical University of South Carolina, Charleston, SC, United States, ²Radiology, University Medical Center Mainz, Mainz, Germany, ³DZHK, Partner-Site Rhine-Main, Mainz, Germany, ⁴Siemens Medical Solutions USA, Inc., Chicago, IL, United States, ⁵Department of Cardiothoracic Surgery, University Medical Center Mainz, Mainz, Germany, ⁶Department of Cardiothoracic Surgery, Medical University of South Carolina, Charleston, SC, United States

This study investigates the effect of the presence of gadolinium contrast on image quality and quantitative flow assessment using conventional and highly accelerated compressed sensing (CS) 4D flow prototypes. In our cohort of 18 healthy volunteers who underwent pre- and post-contrast 4D flow measurements using both techniques, strong to very strong agreement in quantitative flow measurements were found between pre- and post-contrast studies and between conventional and CS 4D flow measurements. SNR was significantly increased after contrast injection in both prototype measurements. Our results suggest that 4D flow acquisition provides accurate flow assessment regardless of the presence of contrast agent.

Judith Zimmermann^1,2, Michael Loecher^1,3, Tyler Cork^1,4, Kathrin Bäumler¹, Alison Marsden^5,6,7, Dominik Fleischmann^1,7, and Daniel Ennis^1,3,7
1Radiology, Stanford University, Stanford, CA, United States, ²Computer Science, Technical University of Munich, Munich, Germany, ³Radiology, Veterans Affairs Health Care System, Palo Alto, CA, United States, ⁴Bioengineering, Stanford University, Palo Alto, CA, United States, ⁵Pediatrics, Stanford University, Stanford, CA, United States, ⁶Bioengineering, Stanford University, Stanford, CA, United States, ⁷Cardiovascular Institute, Stanford University, Stanford, CA, United States

Goal: To assess flow quantitation with multi-directional (ICOSA6) high-moment 4D-flow in a compliant type-B aortic dissection model that provides complex flow and a large velocity range. An in vitro flow setup was designed to acquire prolonged 4D-flow imaging in a flow- and pressure-controlled environment. ICOSA6 4D-flow data was acquired with a highly-sampled stack-of-stars scheme, reconstructed with varying under-sampling factors (R=5-65), and compared to a four-point Cartesian-sampled 4D-flow sequence. Results showed average net flow difference within the ±5% margin for ICOSA6 data under-sampled with up to R=40. Peak velocity, however, differed greatly for all ICOSA6 data when compared to four-point Cartesian.

Aaron Pruitt¹, Yingmin Liu¹, Ning Jin², Peter Speier³, Chong Chen¹, Orlando Simonetti¹, and Rizwan Ahmad^1
1The Ohio State University, Columbus, OH, United States, ²Siemens Medical Solutions USA, Inc., Columbus, OH, United States, ³Siemens Healthcare GmbH, Erlangen, Germany

In this work, we incorporate Pilot Tone as part of our previously described highly accelerated and fully self-gated 4D flow framework to perform retrospective cardiac binning. We compare cardiac triggers derived from Pilot Tone directly with those derived from self-gating and ECG in and demonstrate agreement in aortic and pulmonary artery flow quantification between 4D flow images reconstructed us ECG-, SG-, and PT-based cardiac binning.

Hajime Tamura¹, Hideki Ota², Tatsuo Nagasaka³, Ryuichi Mori³, Chihiro Kato¹, Kohsuke Gonda¹, and Kenichi Funamoto^4
1Department of Medical Physics, Tohoku University, Graduate school of medicine, Sendai, Japan, ²Department of Advanced MRI Collaboration Research, Tohoku University, Graduate school of medicine, Sendai, Japan, ³Department of Radiology, Tohoku University hospital, Sendai, Japan, ⁴Institute of Fluid Science, Tohoku University, Sendai, Japan

We designed a 3-dimensional unicursal channel phantom to simulate small vessels and obtained diffusion-weighted images with varying infusion rate of water. The signal intensities were compared with theoretical data by numerical computation. Our model will allow for understanding the behavior of IVIM images under various flow conditions and evaluating performance of MRI platforms.

Ruponti Nath¹, Sean Callahan¹, Marcus Stoddard², and Amir Amini^1
1ECE, University of Louisville, Louisville, KY, United States, ²Department of Medicine, University of Louisville, Louisville, KY, United States

We propose a novel deep learning-based approach for accelerated 4D Flow MRI by reducing artifact in complex image domain from undersampled k-space. A deep 2D residual attention network is proposed which is trained independently for three velocity-sensitive encoding directions to learn the mapping of complex image from zero-filled  reconstruction to complex image from fully sampled k-space.  Network is trained and tested on 4D flow MRI data of aortic valvular flow in 18 human subjects. Proposed method outperforms state of the art TV regularized reconstruction method and deep learning reconstruction approach by U-net.   

James Rice¹, Labib Shahid¹, Haben Berhane², Joshua Robinson³, Lindsay Griffin³, Cynthia Rigsby³, Michael Markl², and Alejandro Roldan-Alzate^1
1University of Wisconsin-Madison, Madison, WI, United States, ²Northwestern University, Evanston, IL, United States, ³Lurie Children's Hospital, Chicago, IL, United States

4D flow MRI can be used to assess COA repair hemodynamics in-vivo, albeit, after surgery is performed. This study develops a modeling and validation framework for in-vitro COA hemodynamic assessment. Patient-specific in-vitro models were created representing pre- and post- repair COA aortic geometries. 4D flow was employed to assess model hemodynamics. A PIV protocol was developed to assess the validity of the 4D flow MRI results. PIV velocities qualitatively agreed with 4D flow MRI. In-vitro 4D flow MRI can be used as a predictive tool for COA that can be enhanced with PIV validation.

Joao Filipe Fernandes¹, Harminder Gill¹, Julio Sotelo^2,3,4, Shu Wang¹, Alessandro Faraci¹, Cristian Montalba⁵, Jesus Urbina⁶, Ronak Rajani¹, David A. Nordsletten^1,7, Kawal Rhode¹, Sergio Uribe^6,8,9, and Pablo Lamata^1
1School of Biomedical Engineering and Imaging Sciences, King’s College, London, United Kingdom, ²School of Biomedical Engineering, Universidad de Valparaiso, Valparaiso, Chile, ³Biomedical Imaging Center, Pontificia Universidad Católica de Chile, Santiago, Chile, ⁴Millennium Nucleus for Cardiovascular Magnetic Resonance, ANID - Millennium Science Initiative Program, Santiago, Chile, ⁵Biomedical Imaging Center, Pontificia Universidad Catolica de Chile, Santiago, Chile, ⁶Radiology Department, School of Medicine, Pontificia Universidad Catolica de Chile, Santiago, Chile, ⁷Departments Biomedical Engineering and Cardiac Surgery University of Michigan, Ann Arbor, MI, United States, ⁸Biomedical Imaging Center, Pontificia Universidad Catolica de Chile, Santiago, Chile, ⁹Millennium Nucleus for Cardiovascular Magnetic Resonance, Santiago, Chile

Aortic valve (AV) conditions cause extra burden to the heart and frequently lead to clinicalintervention. In the present study we set a 4D-flow-MRI framework to evaluate in-vitro personalizedcompliant 3D-printed AV. We evaluated a healthy and three diseased AV under rest to stresspulsatile flow conditions. The results obtained provide further evidence that trans-valvular non-invasive pressure drop is estimated more accurately accounting for full velocity profile, via thesimplified advective work-energy relative pressure (SAW), than accounting solely for the maximalvelocity as it is clinically stablished. Both the methodology and the findings can potentially improveclinical decision-making.

Kyoung-Jin Park^1,2, Ho-Jin Ha³, Kang-Hyun Ryu⁴, Yang-Dong Hyun², and Dong-Hyun Kim^1
1Electrical & Electronic Engineering, Yonsei University, SEOUL, Korea, Republic of, ²Radiology (Cardiovascular Imaging), University of Ulsan College of Medicine, Asan Medical Center, SEOUL, Korea, Republic of, ³Mechanical and Biomedical Engineering, Kangwon University, Chooncheon, Kangwon-do, Korea, Republic of, ⁴Radiology, Stanford University, Stanford, CA, United States

Compressed sensing (CS) technique has recently been used to accelerated long acquisition time in 4D Flow. However, signal loss from turbulent flow could be problematic when CS method is applied. Signal drop may further aggravate denoising and make error in 4D Flow reconstruction. Using CS technique, we investigated the correlation between velocity error and turbulent flow, TKE estimation error for two motion encoding schemes (i.e., conventional 4D Flow vs ICOSA6).

Ansgar Adler¹, Jost M. Kollmeier², Nick Scholand¹, Sebastian Rosenzweig¹, Yong Wang³, and Martin Uecker^1,4
1Institute for Diagnostic and Interventional Radiology, (UMG) University Medical Center Göttingen, Göttingen, Germany, ²Biomedical NMR, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany, ³Max Planck Institute for Dynamics and Self-Organization, Göttingen, Germany, ⁴DZHK (German Centre for Cardiovascular Research), Göttingen, Germany

The lattice Boltzmann method (LBM) is a versatile numerical technique for simulating complex fluid dynamical systems and beyond. Here, we first describe an extension of the LBM to flow systems in external magnetic fields. The model is verified numerically with several benchmarks and shown to correctly predict signal changes caused by in-flow effects in a simple flow experiment.

Cyril Tous¹, Ivan Dimov¹, Ning Li¹, Simon Lessard¹, and Gilles Soulez^1
1Radiology, Centre de recherche du Centre hospitalier de l’Université de Montréal, Montreal, QC, Canada

4D flow MRI was used to evaluate flow patterns in 35 and 60 degree one bifurcation PVA phantoms (diameter: 4mm main branch, 3mm branches) with and without resistance in one branch. Multiple spatial resolutions were compared with ink injection and manual measurement for validation. Since computational flow models are frequently used in small vessels, 4D flow was compared with a model. Slow, turbulent flow in the outer side of the bifurcation created eddies where some microrobots get trapped. These eddies were quantitatively and qualitatively characterized with 4D flow. Spatial resolution impacts the accuracy of flow measurement at this scale.

Himanshu Singh¹, S Senthil Kumaran¹, and Ganesan Karthikeyan^2
1Department of NMR, All India Institute of Medical Sciences, New Delhi, India, ²Department of Cardiology, All India Institute of Medical Sciences, New Delhi, India

Quantification of blood flow parameter is crucial for assessment of dynamic stability across cardiac vascular system. 4D flow provide systemic estimation of such parameter without requirement of contrast in a single acquisition. Technical limitation of 4D flow restrict dynamic estimation to either heart (LV/RV) or across large vessels due to venc differential. Estimation of large vessel (while retaining the cardiac dynamics) can be assessed accurately with flow estimation using flow velocity and pressure gradient. Choice of parameters is crucial in limited temporal frame using valve/vessel optimised sequence.