Jeudi 12 Octobre

Journée de Formation pluri-professionnelle

Programme organisé par le Groupe de Recherche en Imagerie Cardiaque

(GRIC) Destiné aux Chercheurs et aux Radiologues cardiovasculaires

La thématique de la journée du GRIC qui se tiendra comme chaque année en préambule aux

Journées Françaises de Radiologie (JFR) portera sur l’imagerie du flux sanguin et la biomécanique de

l’aorte. En effet, de récents développements en imagerie médicale permettent d’aborder la

biomécanique de l’aorte (thoracique ou abdominale) avec maintenant des objectifs cliniques

diagnostics ou pronostics. Notamment le développement de l’IRM de flux 4D ouvre de nouvelles

perspectives, et actuellement se développent des systèmes d’imagerie multi-modale et/ou couplée

avec des approches de CFD. Une première session se focalisera sur l’état de l’art sur l’IRM de flux

4D et la biomécanique de l’aorte. Une deuxième session sera dédiée à des communications sur

l’imagerie de flux (quel que soit le type d’imagerie) dans le système cardiovasculaire et sur la

biomécanique de l’aorte. Une troisième session sera consacrée à des communications autour de la

recherche méthodologique en imagerie cardiovasculaire. Cette journée se terminera sur une

discussion autour de la mutualisation des bases de données expertisées en imagerie cardio-

vasculaire. Vous êtes invités à compter de ce jour à soumettre vos résumés sur cette thématique ou

sur tout autre sujet de recherche méthodologique en imagerie cardiaque.

Imagerie du flux sanguin et biomécanique de l'aorte

13h00 - 14h30 : Imagerie 4D du flux sanguin par IRM et biomécanique de l'aorte. Etat de l'art

13h00 Présentation du programme – Alain Lalande

13h05 Biomécanique de l’aorte – Stéphane Avril

The aorta is the most proximal artery connected directly to the heart and acts both as a

conduit and an elastic chamber. In its latter role, the aorta's elasticity serves to convert the

heart's pulsatile flow to nearly steady flow in peripheral vessels. The aorta has received

important attention from the biomechanicists. One of the reasons for the great interest stems

from the observation that increased stiffness of large elastic arteries represents an early risk

factor for cardiovascular diseases. In this talk, we will first review elastic and fracture

properties of the aortic wall and the different techniques used to measure them in vivo or ex

vivo. Average values of important parameters characterizing the biomechanics of the aorta

will be reported. After this review, the talk will focus on important challenges of modern

biomechanics which are to set up predictive and patient-specific numerical simulations. It will

be shown how the knowledge about material properties can now be used for setting up

patient-specific numerical models predicting the structural response of the whole aorta in

different situations (dissection initiation, aneurysm growth and rupture, endovascular

treatment). We will present our recent efforts in measuring regional variations of stiffness

properties across the whole aorta (Fig. 1), especially in our attempt to modelling the growth of

ascending thoracic aortic aneurysms. Future challenges will also be pointed, such as

performing numerical simulations of fenestrated thoracic endograft including fluid-structure

interactions. The question of validating the numerical models in vivo will also be discussed as

bringing confidence into the predictions of FE analysis will be necessary for a future

successful transfer into clinical practice.

Fig. 1. Local stiffness reconstructions across the whole aorta in a patient bearing an aneurysm of the

ascending thoracic aorta

13h30 Principe de l'imagerie de flux par IRM – Aurélien Monnet

Résumé en attente

13h45 Apport d'une technique de flux 4D en pratique clinique – Gilles Soulat

Résumé en attente

14h00 Etude sur fantôme du flux dans l'anévrisme de l'aorte – Alain Lalande

L’anévrisme de l’aorte est une dilatation permanente et localisée d’une artère présentant une

augmentation d’au moins 50% de son diamètre par rapport aux dimensions normales de

l’artère considérée. L'évolution naturelle de l'anévrisme est une augmentation inéluctable de

son calibre suivant la loi de Laplace conduisant à la rupture, inévitable sans prise en charge.

Aujourd’hui, l’intervention chirurgicale est la seule issue pour éviter l’évolution vers la rupture

de l’anévrisme. Les critères utilisés pour évaluer un anévrisme de l’aorte sont basés sur la

mesure de son diamètre mais celle-ci ne s'avère pas toujours très fiables. Il y a un risque

permanent de rupture d'anévrisme quelles que soient ses dimensions et ce, sans symptôme

annonciateur. Aussi, il est nécessaire de développer des protocoles d’aide au diagnostic

basés sur des critères plus proches du comportement physique de l’aorte, en particulier des

critères intégrant les grandeurs de déformations puis de contraintes pariétales ; ces dernières

étant un élément précurseur de la rupture.

Ces paramètres anatomo-fonctionnels représentant les pressions internes ou externes à la

paroi artériel peuvent être déduits de techniques d’imagerie dynamique de type IRM grâce à

une modélisation 4D de l’aorte et une évaluation du flux sanguin au sein de celle-ci.

Actuellement, l’étude des zones de stress, au niveau de la paroi de l’aorte, reste un axe de

recherche où le couplage d’information fluide-solide est jugé pertinent. Elle passe alors par la

validation des modèles numériques proposés pouvant s’appuyer sur l’exploitation de données

issues de l’imagerie sur fantôme de l’aorte. Dans ce cas, ces études sur fantômes requièrent

une installation qui soit compatible avec l’IRM et qui simulent le mouvement cyclique du sang

au sein de l’aorte, avec un flux proche de celui produit lors d’un cycle cardiaque (débit et

pressions). Actuellement se développent des bancs d’expérimentation permettant la gestion

de fantôme au sein de l’IRM avec acquisition dynamique de données (les images ci-dessous

présentent un exemple d’installation au sein du CHU de Dijon). Ces installations comportent

généralement une pompe volumétrique (représentant le cœur) et un circuit hydraulique

(représentant le réseau artériel) sur lequel vient se greffer un fantôme d’AA. En outre, ces

bancs d’expérimentation sont instrumentés (capteurs de pression, débitmètres et chambres

de compliance (en amont et en aval du fantôme)) Ces bancs doivent être relativement

modulables pour accueillir tout type de fantôme. Les fantômes utilisés peuvent être en verre

(rigide) ou en silicone (matériau déformable aux propriétés mécaniques proches de celle des

artères du point de vue macroscopique). Pour être réaliste, le liquide utilisé se doit d’avoir une

viscosité (voire une température) proche de celle du sang. Ainsi, un mélange d’eau et de

glycérine (en proportion dépendant notamment de la température) est une alternative

intéressante au sang.

Le principal avantage de ce type d’expérimentation est de ne pas être contraint par le temps

pour l’acquisition des données et d’offrir une résolution spatio-temporelle optimale des

séquences de flux 4D, l’acquisition de données expérimentales, peu bruitées, étant

primordiale pour étudier de façon précise et pertinente les contraintes mécaniques au sein de

l’aorte. Autre avantage, l’utilisation de fantômes, issus de reconstruction 3D à partir d’images

réalistes, permet d’aborder différents types d’anévrismes de l’aorte thoracique ou abdominale.

L’exposé lors de la journée du GRIC présentera différents aspects techniques de

l’expérimentation de flux 4D sur fantôme de l’aorte ainsi que des résultats que l’on peut

obtenir.

14h15 Flux 4D et mécanique des fluides numériques (CFD) – Ramiro Moreno

Ramiro MORENO1, 2, Thomas PUISEUX2,3, Anou SEWONU1,2, Olivier MEYRIGNAC1, Franck NICOUD3, Simon MENDEZ3, Hervé ROUSSEAU1

1- I2MC, INSERM UMR1048, CHU Rangueil, Toulouse 2- ALARA Expertise, Starsbourg 3- IMAG, CNRS, Montpellier

L'expérience industrielle en modélisation numérique (météo, aéronautique, automobile,

navale) a produit des algorithmes fiables et performants pour résoudre les équations de

mécanique des fluides pour les problèmes complexes. La CFD est devenue la principale

méthode de conception et d'analyse pour ces domaines. Naturellement la médecine s'y est

intéressé pour étudier les flux physiologiques. Les chercheurs physiciens, ingénieurs et

médecins y ont démontré son potentiel dans la compréhension des pathologies cardio-

vasculaires et la mise au point des traitements. Mais les degrés de liberté de cette

technologie sont très importants, ce qui implique qu'il faut maîtriser les conditions du

problème modélisé afin de converger vers la réalité physique, sous peine de créer une

fiction qui illustre une situation complètement irréelle.

De façon indépendante à la CFD, l'imagerie de flux obtenue par IRM s'est développée à

partir d'une approche « monocoupe » vers des acquisitions volumiques « 4D Flow». Le

caractère quantitatif pour cette vélocimétrie par contraste de phase devient alors de plus en

plus strict, et si bien ces examens d'IRM volumique apportent un maximum d’information

hémodynamique, ils nécessitent des temps d’acquisition beaucoup plus longs et font appel

à un nombre croissant de paramètres et traitements pour garantir leur robustesse.

Du point de vue de la physique médicale, plusieurs réflexions se dégagent alors de ces

deux avancées technologiques. Est-ce que les séquences « IRM 4D flow » ont besoin

d'être étalonnées ? De quelle façon la CFD peut contribuer à valider ou enrichir les

mesures obtenues par IRM ? Est-ce que l'IRM peut apporter des mesures pour caractériser

plus précisément les modélisations CFD ? Comment la combinaison des deux méthodes

permet d'avancer sur les versants recherche, développement et clinique ? Dans ce cadre

nous présenterons un dispositif qui permet de faire le parallèle entre les vitesses mesurées

par IRM et modélisées par CFD. Si les résultats de ces deux méthodes sont identiques en

tout point, alors nous avons la maîtrise de l'ensemble des paramètres réels en situation in-

vitro et in-vivo. L'évaluation des écarts représente leur discordance avec la réalité.

Champ de vitesses modélisé (YALES2bio) et mesuré par IRM 4D Flow

14h30 à 14h45 : Pause

14h45 -16h00 : Imagerie du flux sanguin et biomécanique de l'aorte

14h45 Respiratory-resolved self-gated 3D radial 4D flow MRI: Initial results – Monica Sigovan

M. Sigovan1, T. Schneider2, G. Cruz3, C. Mory1, R. Botnar3, L. Boussel 1,4, P. Douek1,4, C. Prieto3 1

University of Lyon, CREATIS Laboratory, Lyon, France 2

Philips Healthcare, Guildford, UK 3

Division of Imaging Sciences and Biomedical Engineering, King’s College London, London, UK 4

Department of Interventional Radiology and Cardio-vascular and Thoracic Diagnostic Imaging, HCL, Lyon, France

Purpose: Develop a respiratory self-gated 3D radial 4D flow MRI sequence and study the

respiratory related variability of flow velocity in the thoracic aorta.

Materials and Methods: A 4D flow MRI sequence based on a spiral phyllotaxis pattern for 3D radial

k- space sampling [1] was implemented on a 1.5T Philips Ingenia system. 4D flow imaging was

performed on 3 healthy volunteers and consisted of a fast field echo interleaved acquisition, 8

projections per interleaf, turbo factor 8, TR/TE 6/2.5 ms, isotropic FOV 340 mm, 2.5 mm isotropic

acquisition voxel, flip angle 6◦, and receiver bandwidth 723 Hz. A total of 54096 radial readouts were

acquired in 6762 interleaves. Each projection was repeated 4 times for velocity measurements using

Hadamard encoding scheme. The respiratory self-gated (SG) signal was derived independently for

each velocity encoding step as the time variation of the z coordinate of the center of mass of the

image [2], computed from the first projection of each interleaf using a conjugate gradient method

with L1 regularization. Using the respiratory SG signal, data was separated in 3 respiratory bins with

equal number of projections. Subsequently, each respiratory phase was binned in 8 cardiac phases

using the ECG signal. Respiratory- and cardiac-resolved images were reconstructed offline using a

standard gridding algorithm and velocity images were computed using complex phase subtraction.

Blood flow velocities were measured at different locations in the thoracic aorta.

Results: All datasets were reconstructed successfully with minimal residual breathing artefacts.

Respiratory amplitude of around 1 cm was observed. Time average velocity values in the thoracic

aortas of the three volunteers varied between respiratory phases, from 59 ± 14 cm/s in phase 1

(corresponding to inspiration), to 57 ± 10 cm/s in phase 2, and 49 ± 5 cm/s in phase 3 (expiration).

Discussion: We presented a respiratory and cardiac resolved 4D Flow MRI sequence based on a

3D radial trajectory. Our preliminary results highlighted respiratory cycle related velocity variations,

but will have to be confirmed with a Cartesian 4D flow sequence. This work is ongoing.

References:

[1] D. Piccini et al, Magnetic Resonance in Medicine 66:1049–1056 (2011)

[2] O. Wieben et al, Proc. Intl. Soc. Mag. Reson. Med 9 (2001) #737

15h00 The Assessment of Aortic Pulse Wave Velocity Using 4D Flow Magnetic Resonance

Imaging: Methods Comparison – Sophia Houriez-Gombaud-Saintonge

Sophia Houriez--Gombaud-Saintonge1,2, Elie Mousseaux3, Ioannis Bargiotas1, Alain De Cesare1, Thomas Dietenbeck1, Kévin Bouaou1, Alban Redheuil1, Gilles Soulat3, Umit Gencer3, Damian Craiem4, Yasmina Chenoune2, Nadjia Kachenoura1

1Sorbonne Universités, UPMC Univ Paris 06, INSERM, CNRS, Laboratoire d’Imagerie Biomédicale, Paris, France 2ESME SUDRIA, ESME Research Lab, PARIS, France 3Hopital Européen Georges Pompidou, PARIS, France 4 Universidad Favaloro-CONICET, IMeTTyB, Buenos Aires, Argentina

Aim. Arterial pulse wave velocity (PWV) is associated with increased CV mortality in aging and disease. Several studies have shown the accuracy of aortic PWV (aoPWV) estimation using 2D+t MRI (aoPWV=distance between two aortic locations/transit-time (TT) needed by flow wave to travel between these locations). Our primary aim was to estimate aoPWV from 4DFlow MRI, which has low resolution counterbalanced by large anatomical coverage. Specifically we aim to compare various methods of TT and consequently aoPWV estimation in terms of associations with age and Bramwell-Hill (BH) aoPWV estimated from aortic distensibility.

Method We studied 43 healthy subjects (48±17years) who had MRI including aortic 4DFlow. Aorta was semi-automatically segmented from angiograms estimated as a time-average of frames combining anatomical modulus with velocities llowingflow-rate curves estimation in plans perpendicular to aortic centreline. Three strategies were used to estimate aoPWV: (S1) using flow curves in two aortic locations (ascending (AA) and distal descending aorta (dDA)) is a 2D-like strategy, (S2 and S3) using flow curves of the entire aortic path-line between dDA and AA. For S1, TT was calculated using 3 approaches: cross-correlation (time domain-TTC), wavelet (time-frequency domain-TTW) and Fourier (frequency domain-TTF). For S2, such TT estimates were used iteratively from dDA to AA resulting in distance-time dot-plot and its slope provided aoPWV. For S3, a planar fit of all flow curves up-slope resulted in aoPWV.

Figure 1. Aortic 4D flow image with plan positioning

Results. Expected associations with age were found for the three strategies with strongest correlations for 3D-like strategies (S2:r=0.75, S3: r=0.72,p<0.001), compared to 2D-like strategy (S1: r=0.55,p<0.001). Similar results were found for associations with BH aoPWV. Independent of the strategy, best results were obtained using the wavelet-based approach, as compared to time domain and Fourier approaches.

Conclusion. The loss in temporal and spatial resolutions in 4DFlow data is fully compensated by aortic 3D coverage, leading to stronger associations with age and aortic distensibility for the 4DFlow aoPWV strategies taking advantage of the full 3D coverage.

Linear coefficient of regression (R)

R

S1 S2 S3 TTc TTw TTf TTc TTw TTf

15h15 Patient-specific numerical study of hemodynamic pattern associated with valve anatomy

– Clément Acquitter

Clément Acquitter1,2,4, Masato Ogitsu2, Stéphanie Bricq1, Jean-Joseph Christophe4, Olivier Bouchot3, Suguru Miyauchi2, Toshiyuki Hayase2, Alain Lalande1,3

1Le2i, Université de Bourgogne, Allée Alain Savary, Dijon, France 2Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, 980-0812, Japon

3CHU de Dijon, 1 Bld Jeanne d’Arc, BP 77908, 21079 Dijon, France 4CASIS – Cardiac Simulation & Imaging Software, 64A rue Sully, 21000 Dijon, France

ABSTRACT Bicuspid aortic valve (BAV) is the most common congenital heart disease and is associated with an increased prevalence of aortic aneurysm. In this study, we have used a patient-specific numerical model of the aorta to investigate the relation between valve morphology and ascending aortic aneurysm (AAA). We have compared ascending aortic hemodynamics in patients with tricuspid and bicuspid valve. Results demonstrate altered hemodynamics and high WSS areas at the level of the ascending aorta in patient with a BAV.

1. Introduction Bicuspid aortic valve (BAV) is the most common

congenital heart disease and is associated with an increased prevalence of aortic aneurysm [1]. In the

present study, we proposed a patient-specific numerical model of the thoracic aorta which allows to investigate flow patterns and stress areas associated with valve anatomy. Our methodology combine 4D Flow MRI and Computational Fluid Dynamics (CFD) to study hemodynamic patterns associated with valve anatomy.

2. Method

Image acquisition Data were acquired on a 1.5T magnet (Siemens) with a 4D flow-sensitive MRI sequence on three patients (patient 1: tricuspid aortic valve (TAV), patient 2: BAV+ ascending aortic aneurysm (AAA), patient 3: BAV).

Geometry reconstruction Image segmentation has been performed on the phase-contrast MR angiography (PCMRA) image [2].

Numerical simulation CFD simulations were

performed on fine tetrahedral meshes over three cardiac cycles. The velocity profile extracted from the 4D Flow MRI at the level of the valve is used as inlet boundary condition (BC).

3. Results and Discussion Flow analysis Patients with BAV show flow

separation with high velocity asymmetric jet flow and recirculating vortex. On the other hand, TAV patient shows a higher inflow homogeneity (Fig. 1)

Fig. 1 Flow patterns at peak systole. (1: TAV, 2: BAV+AAA, 3: BAV)

WSS analysis Patients with BAV exhibit higher WSS areas at the level of the ascending aorta compared to the tricuspid case (Fig. 2). These results corroborate with previous study [4,5] and demonstrate that the asymmetry of the jet flow for BAV patients has an important impact on stress areas at the level of the ascending aorta.

Fig. 2 WSS analysis right after peak systole. (1: TAV, 2: BAV+AAA, 3: BAV)

4. Concluding Remarks We have proposed a patient-specific numerical model of the thoracic aorta where BC at the inlet are extracted

from 4D Flow MRI. We have used this model on three patients among which two have a bicuspid valve. These results can contribute to explore the relation between valve geometry and AAA.

References [1] D. Lavall et al., Deutsches Ärzteblatt International, 109 (2012). [2] B. Köhler et al., Computer Graphics Forum, (2016). [3] M. Sigovan et al., Journal of Magnetic Resonance Imaging, 34 (2011), 1226-1230. [4] E. Faggiano et al., Biomechanics and Modeling in Mechanobiology, 12 (2013), 801-813. [5] F. Viscardi et al., Artificial Organs, 34 (2010), [6] 1114– 1120.

15H30 Relative Aortic Blood Pressures Using 4D flow MRI: Associations with Age and Aortic

Tapering – Kevin Bouaou

Kevin Bouaou1, Ioannis Bargiotas1, Damian Craiem3, Gilles Soulat2, Thomas Dietenbeck1, Sophia Houriez-Gombaud-Saintonge1, Alain De Cesare1, Umit Gencer2, Alain Giron1, Alban Redheuil1, Didier Lucor4, Elie Mousseaux2, Nadjia Kachenoura1.

1Sorbonne Universités, UPMC Univ Paris 06, INSERM 1146, CNRS 7371, Laboratoire d’Imagerie Biomédicale, Paris, France 2Hôpital Européen Georges Pompidou, Paris, France 3Universidad Favaloro-CONICET, IMeTTyB, Buenos Aires, Argentina 4LIMSI, CNRS, Université Paris-Saclay, Orsay, France

Aims: Aortic pressure gradients are useful in characterizing diseases such as valvular stenosis and aortic

coarctation. Although catheterization is the gold standard for the measurement of local aortic pressures, its invasiveness limits its usefulness in clinical routine. Alternatively, applanation tonometry has been proposed for a non-invasive evaluation of central arterial pressure variations through time. Although it is well accepted that pressure distribution varies locally throughout the aorta, none of the aforementioned methods can provide the exact spatial pressures distribution. Accordingly, our aim was to use 4D flow MRI with Navier-Stokes equations to generate volumetric relative pressures (3D+t) and to assess: 1) relationship between trans-aortic pressure gradient with age and aortic tapering (proximal to distal change in lumen area), 2) effect of temporal resolution.

Methods: We studied 47 healthy subjects (49±17.6 years) who underwent 4D flow MRI with 50 phases per cardiac cycle. Among them 20 were reconstructed with 20 phases. Spatio-temporal pressure maps were computed by: 1) spatially integrating pressure gradients obtained from Navier-Stokes equations, while assuming zero pressure at the level of aortic valve, 2) applying an iterative refinement resulting in smooth relative pressures within the segmented aorta through time. For all datasets, relative pressure was averaged in six aortic sections perpendicular to aortic centerline and covering both proximal (PA) and distal aorta (DA).

Results: DA to PA pressure gradient decreased with age (r=0.60, p<0.05), and was inversely related to DA/PA areas ratio (r=0.35, p<0.05). This is in line with the physiological evidence indicating that in aging DA area tends to equalize with PA area lowering the DA to PA pressure gradient. Finally, peak systolic pressure was higher when considering the 50 phases data.

Conclusion: Relative pressures calculated from 4D flow MRI are consistent with prior physiological knowledge as demonstrated by their variations with age and with aortic geometry.

15h45 in vitro Validation of 1D Blood Flow Model and Application on Realistic Aorta Phantom –

Khalid Rachid

Khalil Rachid1, Astrid Decoene2, Dima Rodriguez1

1Laboratoire d’Imagerie par Résonance Magnétique Médicale et Multi-Modalités, Univ. Paris-Sud, CNRS, Université Paris-

Saclay, Orsay, France; 2Laboratoire de Mathématiques d’Orsay, Univ. Paris-Sud, CNRS, Université Paris-Saclay, Orsay,

France.

INTRODUCTION

Central aortic blood pressure (ABP) is known to provide valuable information on the

biomechanical properties of large elastic vessels as well as on the left ventricle function [Shi et

al, 2011]. Moreover, ABP - rather than peripheral BP - is strongly correlated to adverse

cardiovascular outcomes [McEniery et al, 2014]. However, the invasive nature of ABP standard

measurements still limits its clinical applica- tions. The aim of this work is to non-invasively

estimate an absolute ABP using a 1D blood flow model coupled to Magnetic Resonance (MR)

derived data. First, we validate our model on a straight elastic tube, then we assess its

applicability to a realistic aorta phantom. Additionally, we present here an MR-experimental

evaluation of the hemodynamic effects induced by a localized stiffness.

MATERIAL & METHODS

The 1D model: derived from the integration of the Navier-Stokes equations over a cross-sectional area of a straight compliant tube. A first order finite volume scheme was written in Python to solve numerically the hyperbolic system [Delestre et al, 2015]. An MR- derived flow waveform was applied to the model as an inlet boundary condition.

Experimental Setup: consists of a pulsatile flow pump that carries the blood mimicking fluid through the phantoms. Measured pressure waves at two phantom proximal and distal sites were compared to the model predictions at the same anatomical locations.

Figure 1. (a) Experimental setup; (b) The elastic straight tube; & (c) The realistic aorta phantom used in this study.

Local Stiffening: Rigid pieces, well adapted to the phan- tom geometry at 4 aorta sites

(Fig.2) were 3D printed. When positioned at the correspond- ing site they reduce its dis-

tension and simulate a local- ized stiffness. Induced hemo- dynamic changes were then

compared to those of the con- trol case.

Figure 2. 4 sites stiffening of the aorta phantom

RESULTS

1D Model validation:

Figure 3. Simulated versus measured proximal (a) and distal pressure (b) in the compliant straight tube.

Figure 4. Simulated versus measured proximal (c) and distal pressure (d) in the realistic aorta phantom.

Straight tube Aorta phantom

Predicted

Measured

Predicted

Measured

S.P (mmHg) 261.65 255.76 133.98 133.01

D.P (mmHg) 60.00 51.81 64.08 63.11

P.P (mmHg) 201.65 203.95 69.90 69.90

RMSE (%) 6.26 4.99

Table I

Systolic (S.P), diastolic (D.P) and pulse pressure (P.P) of simulated and measured proximal pressure. RMSE = root mean square error.

Localized stiffness effect on compliance :

Figure 5. Estimated local compliance of the proximal site upstream to all stiffness configurations.

DISCUSSION

We validated a 1D flow model applied to a straight compliant vessel. We also showed that,

although aortic curvature is not considered in the model formulation, its application to the

realistic aorta phantom gives satisfactory results. On the other hand, the expected stiffness-

induced alterations on pressure and flow rate have been observed. Moreover, we noted that a

localized stiffness extends its effects to its adjacent sites by reducing the upstream site

compliance.

16h00 à 16h15 : Pause

16h15 -17h30 : Recherche méthodologique en imagerie cardiaque

16h15 Combined imaging of myocardial metabolism and tissue stiffness using Positron

emission tomography and ultrafast ultrasound after myocardial infarction in the rodent

heart – Joevin Sourdon

Joevin Sourdon1, Béatrice Berthon2, Anikitos Garofalakis1, Xumeng Zhang1, Thomas

Viel1, Mickaël Tanter2, Jean Provost2, Bertrand Tavitian1,3. 1 Paris Cardiovascular Research Center (PARCC); INSERM UMR970; Université Paris Descartes; Paris, France 2 Institut Langevin, ESPCI Paris, PSL Research University, CNRS UMR 7587, INSERM

U979 3 Service de Radiologie, Hôpital Européen George Pompidou, Assistance Publique-Hôpitaux de Paris (AP-HP), Paris, France

Objective: To obtain coregistered images of myocardial metabolism using Positron emission tomography (PET), and tissue perfusion and tissue stiffness using ultrafast ultrasound imaging (UUI) in a rat model of myocardial infarction. Background: After an acute myocardial ischemic event, identification of viable / non-viable tissue is useful to define treatment. PET reports on myocardial glucose consumption. Shear- wave imaging (SWI) measures myocardial stiffness. Simultaneous imaging of both parameters should improve the confidence in viability prediction. Methods: Simultaneous PET-UUI was performed in 6 rats before (baseline) and 7 days after ligation of the left anterior descending coronary artery. During the same imaging session, cardiac metabolism was assessed after injection of 2-deoxy-

2[18F]fluoro-D-glucose (FDG), and myocardial stiffness with SWI. Image volumes were co-registered and segmented using original software, and analyzed using Amira® and PMOD®. Results: One week after the onset of permanent ischemia, the infarcted zone showed a significantly reduced standard uptake value (SUV) of FDG compared to baseline (3.90±0.48 versus 6.32±0.83; p<0.01), while the border zone had preserved SUV (6.27±0.65). SW velocity representing tissue stiffness was similar in the infarcted area and in the border zone, and in both regions was significantly higher than at baseline (2.50±0.22 and 2.35±0.15 versus 1.62±0.08, respectively; p<0.05). Conclusions: Multi-parametric imaging of the ischemic rodent heart identifies distinct areas in the injured myocardium. This preliminary study supports the hypothesis that simultaneous co- registered PET-UUI can distinguish viable from non-viable myocardium and could become a useful tool for defining therapeutic options.

16h30 Comparison between invasive and non-invasive estimation of left ventricular

pressure-strain loops – Arnaud Hubert

Arnaud HUBERT, Virginie LE ROLLE, Elena GALLI, Alfredo HERNANDEZ, Erwan

DONAL,

Laboratoire du Traitement du Signal et de l'image (unité INSERM U1099, université de Rennes1), Service de Cardiologie et Médecine vasculaire et CIC-IT 1414 (CHU de Rennes).

Purpose: The area of Left ventricular (LV) pressure-strain loop (PSL) is used as an index of regional cardiac work, as it has shown a high correlation with regional metabolic activity. This marker might be useful to better select cardiac resynchronization therapy (CRT) responders. Russel et al. recently proposed a non-invasive method to estimate LV PSL area based on a standard waveform that was fitted to valvular events and scaled to systolic blood pressure. The purpose of this work is to compare the invasive versus the non-invasive estimation of LV PSL areas.

Methods: In five patients recently implanted with a CRT, LV pressure was invasively measured in five conditions: CRT-off, LV pacing, right ventricular (RV) pacing and two different biventricular (BIV) pacing configurations. For each condition, systolic blood pressure was also measured non-invasively, by a brachial artery cuff. Trans-thoracic echocardiography (TTE) loops were recorded simultaneously. LV strain was calculated in off-line analysis. The relative root mean square error (rRMSE) and the linear correlation coefficient (R2) between invasive and non-invasive estimated PSL areas were calculated for each patient and each configuration. A Bland-Altman (BA) analysis was also performed.

Results: 23 different haemodynamic conditions were compared, for 414 different segmental PSL. The rRMSE between non-invasively and invasively estimated LV pressures was between 5.8 and 13.8 mmHg. The invasively and non-invasively estimated LV-PSL were strongly correlated, with an R² always equal or higher than 0.97 (Table1). The global correlation was R²=0.977, p<0.001 (figure 1A). BA analysis (Figure 1B) shows that the mean bias for the estimation of segmental LV-PSL area is 52.8 (mean area736 vs 789 mmHg.%, for non-invasive and invasive LV-PSL respectively). A significant bias effect with linearly increasing error with pressure values was also observed in Figure 2B.

Conclusion: The non-invasive estimation of LV pressure-strain loop area strongly correlates with invasive measurements. However, a linear bias error with increasing pressure values has been observed using a BA analysis.

Table 1: Comparison between estimated and measured left ventricular pressure and pressure-

strain loops which as a result.

CRT off RV pacing LV pacing BIV pacing 1 BIV pacing 2

rRMSE (mmHg)

PSL (R2) rRMSE (mmHg)

PSL (R2) rRMSE (mmHg)

PSL (R2) rRMSE (mmHg)

PSL (R2) rRMSE (mmHg)

PSL (R2)

Patient1 13.8 0.99 13.0 0.99 10.4 0.98 11.2 0.99 - -

Patient2 9.8 0.99 9.2 0.97 10.5 0.99 8.9 0.99 10.7 0.98

Patient3 12.3 0.99 10.3 0.99 - - 11.6 0.99 11.6 0.99

Patient4 8.8 0.98 6.7 0.99 9.5 0.99 6.8 0.97 13.7 0.99

Patient

5 7.9 0.99 5.8 0.99 7.9 0.99 6.4 0.99 7.4 0.99

PME: pressure mean error; PSL: Pressure strain loop area; rRMSE: relative root mean square

error

Figure1: Panel A: Correlation between estimated and measured segmental LV-PSL area. Panel

B: Bland and Altman graph of estimated and measured segmental LV-PSL.

16h45 Réserve élastique de l’aorte thoracique et abdominale à l’effort et Marfan –

Laurence Bal

Laurence BAL1,2, Alain LALANDE3, Monique BERNARD1, Alexis

JACQUIER1,4

1 CRMBM, UMR 6612 CNRS-Université de la Méditerranée, Medecine University of Marseille 2 Aortic Center,Department of Vascular Surgery and Vascular Medecine, La Timone Hospital,

Marseille 3 LE2I, University of Burgundy and MRI department, University Hospital of Dijon, Dijon 4 Department of Radiology and Cardiovascular Imaging, La Timone Hospital, Marseille

Aim. Evaluation of the aortic “elastic reserve” might be a relevant marker to assess the

risk of aortic event, in addition to aortic dilatation. Our aim was to compare regional aortic

elasticity at rest and during supine bicycle exercise at 1.5 T MRI in patient with Marfan

syndrome.

Methods. Eleven patients with Marfan syndrome (mutation in FBN1) completed the

entire protocol. Images were acquired immediately following maximal exercise.

Retrospective cine sequences were acquired before and after exercise to assess

compliance and distensibility at four different locations: ascending aorta (AA), proximal

descending aorta (PDA), distal descending aorta (DDA) and aorta above the coeliac

trunk level (CA) [1]. The patients made an effort during the examination such as to obtain

twice the resting heart rate. The aortic compliance and distensibility were automatically

calculated with a dedicated software [2].

Results. Exercise induced a decrease of aortic compliance and distensibility as shown in

the following table. This decrease is statistically significant for the AA, the PDA and for

the compliance for DDA.

Aortic compliance (mm²/mmHg) Aortic distensibility (mmHg-1 × 103)

rest exercise p rest exercise p

AA 2.81 ±

0.85

1.63 ±

0.79 0.004

5.76 ±

2.70 2.84 ± 1.66 0.011

PDA 1.60 ±

0.42

1.01 ±

0.53 0.010

5.99 ±

1.93 3.70 ± 1.91 0.016

DDA 1.69 ±

0.57

1.08 ±

0.61 0.024

6.97 ±

1.87 6.73 ± 2.14 0.824

CA 1.64 ±

0.68

1.32 ±

0.62 0.331

8.11 ±

3.46 5.35 ± 2.37 0.168

Discussion. This preliminary study shows that assessment of regional aortic function for

patients with Marfan syndrome during exercise is feasible using MRI. During stress,

aortic elasticity decreases significantly. However, equivalent results were ever found for

healthy subjects [1]. Further studies are required to compare these results with normal

subjects with matched sex and age, and to create thresholds (“z-score of distensibility”)

for ascending aorta dysfunction among patients with Marfan or related diseases.

[1] Bal et al. PLoS One. 2016; 11(6). [2] Rose et al. Magn Reson Imag. 2010; 28: 255-263.

17h00 Étude de l'évolution des zones de stagnation du sang dans les anévrismes de

l'aorte abdominale – Florian Joly

F. Joly (Montréal, Canada), S. Lessard (Montréal, Canada), C. Kauffmann (Montréal, Canada),

G. Soulez (Montréal, Canada)

Objectifs : La mesure du diamètre maximal reste la référence pour le suivi de l’anévrisme de

l'aorte abdominale AAA et la planification interventionnelle. Cependant, la prédiction de la

croissance et l’évaluation du risque associé reste un défi majeur.

Notre objectif est la quantification de stagnation du sang dans l'AAA, liée au dépôt de thrombus

lié à la dégradation prématurée de la paroi aortique et la croissance rapide de l'AAA.

Matériel et méthode :

- Segmentation des AAAs de 9 scans de suivis pré-opératoire d'un patient

- Calcul numérique de l’écoulement, validé par IRM

- Extraction des zones de stagnations dynamiques à partir des Structures Lagrangiennes

Cohérentes (SLC)

Résultats : Pour chaque scans, 2 zones de stagnations étaient visibles. Au cours de la

croissance de l'AAA, leur volume a augmenté de 127% et 293%. La partie stagnante de la

lumière est passée de 25% à 50% de son volume total en 7 ans. Les volumes de stagnation ont

évolué de manière continue et quasi linéaire (r^2= 0.98 et 0.84).

Conclusion : Nous proposons un outil robuste et innovant permettant de quantifier les zones de

stagnation du sang dans l’AAA, habituellement inaccessible dans un écoulement pulsé. Ces

résultats pourront jouer un rôle majeur pour la planification interventionnelle et la gestion du

risque personnalisé.

Vue des 2 zones de stagnation visibles dans un anévrisme au cours de sa croissance, extraites

du champ d'écoulement calculé pour les années 2006, 2009 et 2012.

17h15 Characterization of the physiological displacement of the aortic arch using non-

rigid registration and MR imaging – Bahaa Nasr

Bahaa Nasr, MD a,b,c; Florent Le Ven, MD c,d; Joel Savean, MS b; Douraied Ben Salem, MD, PhD c,e; Michel Nonent, MD, PhD c,e; Pierre Gouny, MD, PhD a,c; Dimitris Visvikis, PhD b; Hadi Fayad, PhD b,c.

a Service de Chirurgie Vasculaire, CHU Brest, Brest, F-29200 France

b INSERM UMR 1101, LaTIM, CHU Morvan, Brest, F-29200 France

c Université de Bretagne Occidentale, Brest, F-29200 France

d Service de cardiologie, CHU Brest, Brest, F-29200 France

e Service de radiologie, CHU Brest, Brest, F-29200 France

Abstract (255)

Objectives: The aim of this work was to study the physiological aortic arch three-dimensional displacement using non-rigid registration methods and magnetic resonance imaging (MRI).

Materials/Methods: Ten healthy volunteers underwent thoracic magnetic resonance imaging. A

prospective cardiac gating was performed with a 3D turbo field echo sequence in order to obtain end-systolic and end-diastolic MR images. The rigid and elastic behavior between these two cardiac phases was detected and compared using either an affine or an elastic registration method. To assess reproducibility, a second MRI acquisition was performed 14 days later.

Results: Affine registration between the end-systolic and end-diastolic MR images showed significant global translations of the aortic arch and the supra-aortic vessels in the x, y and z directions (2.02 ± 1.6, -0.71 ± 1.1 and -1.21 ± 1.4 mm respectively). Corresponding elastic registration indicated significant local displacement with a vector magnitude of 5.1 ± 0.89 mm for the brachiocephalic artery (BCA), of 4.26 ± 0.83 mm for the left common carotid artery (LCCA) and of 4.8 ± 0.86 mm for the left subclavian artery (LSCA). There was a difference in displacement between the supra-aortic trunks of the order of 2 mm. vectors displacement were not statistically different between the repeated acquisitions.

Conclusions: Our results showed important deformations in the ostia of supra-aortic vessels

during cardiac cycle. Aortic arch motions seems to be an additional factor to take into account

when designing and manufacturing fenestrated endografts. The elastic registration method

seems to provide more precise results, but is more complex and time consuming

17h30 -18h00 : Discussion autour de la mutualisation des bases de données expertisées en

imagerie cardio-vasculaire

17h30 Présentation du projet Human Heart Project

Contacts : Mr Alain Lalande Alain.Lalande@u-bourgogne.fr Pr Alexis Jacquier Alexis.JACQUIER@ap-hm.fr