emerging trends in cv flow visualizationunina.stidue.net/universita' di trieste/ingegneria...
TRANSCRIPT
doi:10.1016/j.jcmg.2012.01.003 2012;5;305-316 J. Am. Coll. Cardiol. Img.
Ebbers, Giovanni Tonti, Alan G. Fraser, and Jagat Narula Partho P. Sengupta, Gianni Pedrizzetti, Philip J. Kilner, Arash Kheradvar, Tino
Emerging Trends in CV Flow Visualization
This information is current as of March 19, 2012
http://imaging.onlinejacc.org/cgi/content/full/5/3/305located on the World Wide Web at:
The online version of this article, along with updated information and services, is
by on March 19, 2012 imaging.onlinejacc.orgDownloaded from
S
EP
J
C
Tai
AT(fp
tiw
MCT1I2
3
45
F†CttHst
M
J A C C : C A R D I O V A S C U L A R I M A G I N G V O L . 5 , N O . 3 , 2 0 1 2
© 2 0 1 2 B Y T H E A M E R I C A N C O L L E G E O F C A R D I O L O G Y F O U N D A T I O N I S S N 1 9 3 6 - 8 7 8 X / $ 3 6 . 0 0
P U B L I S H E D B Y E L S E V I E R I N C . D O I : 1 0 . 1 0 1 6 / j . j c m g . 2 0 1 2 . 0 1 . 0 0 3
T A T E - O F - T H E - A R T P A P E R
merging Trends in CV Flow Visualizationartho P. Sengupta, MD, DM,* Gianni Pedrizzetti, PHD,† Philip J. Kilner, MD‡
Arash Kheradvar, MD, PHD,§ Tino Ebbers, PHD,� Giovanni Tonti, MD,¶Alan G. Fraser, MB,# Jagat Narula, MD, PHD*New York, New York; Trieste, Italy; London, United Kingdom; Irvine, California; Linköping,
Sweden; Sulmona (AQ), Italy; and Cardiff, United KingdomCaredmtpto
Cap
ABvaCros
Ml
C
I
ACC: CARDIOVASCULAR IMAGING CME
ME Editor: Ragaven Baliga, MD
his article has been selected as this issue’s CME activity,vailable online at www.imaging.onlinejacc.org by select-ng the CME tab on the top navigation bar.
ccreditation and Designation Statementhe American College of Cardiology Foundation
ACCF) is accredited by the Accreditation Councilor Continuing Medical Education (ACCME) torovide continuing medical education for physicians.
The ACCF designates this Journal-based CME ac-ivity for a maximum of 1 AMA PRA Category 1 Cred-t(s)™. Physicians should only claim credit commensurateith the extent of their participation in the activity.
ethod of Participation and Receipt of CMEertificateo obtain credit for this CME activity, you must:. Be an ACC member or JACC: Cardiovascularmaging subscriber. Carefully read the CME-designated article avail-
able online and in this issue of the journal.. Answer the post-test questions. At least two out
of the three questions provided must be answeredcorrectly to obtain CME credit.
. Complete a brief evaluation.
. Claim your CME credit and receive your certif-icate electronically by following the instructions
anuscript received November 29, 2011; revised manuscript received
imaging.onlinejacc.oDownloaded from
ME Objective for This Article: At the end of thisctivity the reader should be able to: 1) understandelevant concepts in cardiac fluid mechanics, including themergence of rotation in flow and the variables thatelineate vortical structures; 2) elaborate on the mainethods developed to image and visualize multidirec-
ional cardiovascular flow; 3) develop dedicated imagingrotocols for particle imaging velocimetry; and 4) discusshe potential clinical applications and technical challengesf determining multidirectional cardiovascular flow.
ME Editor Disclosure: JACC: Cardiovascular Im-ging CME Editor Ragaven Baliga, MD, has re-orted that he has no relationships to disclose.
uthor Disclosure: Dr. Kilner is supported by theritish Heart Foundation and by the NIHR Cardio-ascular Biomedical Research Unit of Royal Bromptonnd Harefield NHS Foundation Trust and Imperialollege, London, United Kingdom. Dr. Fraser has
eceived research support from Hitachi-Aloka. Allther authors have reported that they have no relation-hips relevant to the contents of this paper to disclose.
edium of Participation: Print (article only); on-ine (article and quiz).
ME Term of Approval:
ssue Date: March, 2012
given at the conclusion of the activity. Expiration Date: February 28, 2013rom the *Zena and Michael A. Wiener Cardiovascular Institute, Mount Sinai School of Medicine, New York, New York;University of Trieste, Trieste, Italy; ‡Royal Brompton Hospital and Imperial College, London, United Kingdom; §University ofalifornia Irvine, Irvine, California; �Linköping University, Linköping, Sweden; ¶SS Annunziata Hospital, Sulmona (AQ), Italy; and
he #Wales Heart Research Institute, School of Medicine, Cardiff University, Cardiff, United Kingdom. Dr. Kilner is supported byhe British Heart Foundation and by the NIHR Cardiovascular Biomedical Research Unit of Royal Brompton and
arefield NHS Foundation Trust and Imperial College, London, United Kingdom. Dr. Fraser has received researchupport from Hitachi-Aloka. All other authors have reported that they have no relationships relevant to the contents ofhis paper to disclose.
January 6, 2012, accepted January 9, 2012.
by on March 19, 2012 rg
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
306
Emerging Trends in CV Flow Visualization
Blood flow patterns are closely linked to the morphology and function of the cardiovascular system. These patterns
reflect the exceptional adaptability of the cardiovascular system to maintain normal blood circulation under a wide range
of workloads. Accurate retrieval and display of flow-related information remains a challenge because of the processes
involved in mapping the flow velocity fields within specific chambers of the heart. We review the potentials and pitfalls
of current approaches for blood flow visualization, with an emphasis on acquisition, display, and analysis of multidirec-
tional flow. This document is divided into 3 sections. First, we provide a descriptive outline of the relevant concepts in
cardiac fluid mechanics, including the emergence of rotation in flow and the variables that delineate vortical structures.
Second, we elaborate on the main methods developed to image and visualize multidirectional cardiovascular flow,
which are mainly based on cardiac magnetic resonance, ultrasound Doppler, and contrast particle imaging velocimetry,
with recommendations for developing dedicated imaging protocols. Finally, we discuss the potential clinical applica-
tions and technical challenges with suggestions for further investigations. (J Am Coll Cardiol Img 2012;5:305–16)
© 2012 by the American College of Cardiology Foundation
d
Flow through the heart interacts with the mobilecontours of the myocardium, valve, and vessels.Blood flowing through the sequence of compart-ments is subject to changes in direction and luminaldiameter, and as a consequence, flow is multidirec-tional and vortical with a tendency to curl or spin inthe cardiac chambers during various phases of thecardiac cycle (1). The valves, chamber geometry,and wall motion modify the flow patterns to pro-duce a hemodynamic environment comprising nor-mal or pathological adaptation. Therefore, analyz-ing the spatial and temporal distribution of bloodflow in the cardiovascular system may providediagnostic and prognostic information. However,acquiring and visualizing flow through the vol-umes of the heart cavity is a complex topic.Doppler echocardiography and cardiac magneticresonance (CMR), in different ways, have beenused to measure the velocities of the flow that ispredominantly unidirectional while it passesthrough a valve, a jet, or an approximately cylin-drical vessel segment. Recent technological inno-vations in imaging modalities and the emergenceof flow visualization techniques have providedvaluable opportunities for direct in vivo assess-ment of multidirectional blood flow. Analysis ofthe flow fields is associated with variables such aspressure differences, shear stresses, and energydissipation. The display and measurement of suchflow variables on a page or 2-dimensional (2D)computer screen can be challenging, and eachvariable calls for different types of analysis andrepresentation. Therefore, there is a pressing
need to guide and advance flow visualizationimagiDownloaded from
techniques toward the development of consistentmeasurements and clinical applications (2).
Basic Principles in Cardiovascular Fluid Mechanics
Blood is a corpuscular material that behaves like afluid and can be deformed throughout its volume.Its movements are powered and constrained by themovements of containing boundaries. Blood’s pat-terns of flow are subject, on the one hand, to thedirectional momentum of the streams entering andmoving through the chambers and, on the otherhand, to the slowing, frictional effect of the viscosityboth relative to the boundaries as well as betweenstreams, with different velocities sliding against oneanother within the volume. The liquid motion canbe imagined as being composed of numerous flex-ible layers (laminae) sliding relative to each other(Fig. 1A).
Laminar flow, in which the directions of stream-lines remain almost parallel to those of neighboringstreamlines or boundaries, occurs in the smaller,more peripheral blood vessels, but it is not typical offlow through large vessels or the chambers of theheart. Laminar flow tends to become unstable,fragmenting and mixing with eddies and counter-eddies when its kinetic energy (which is propor-tional to the square of its velocity [V]) becomeslarge relative to its rate of energy dissipation (whichis roughly proportional to �V/D, where D is the
iameter of the vessel and � is the kinematicviscosity). When this ratio, known as the Reynoldsnumber (Re � VD/�), exceeds a critical value
(approximately 2,500 for steady flow in a straightby on March 19, 2012 ng.onlinejacc.org
eb
tgprmcu(tcivc3vvtCpttCs
hastu(bpfiaflkflim
vector flow mapping
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
307
circular vessel), laminar motion becomes unstable,and the fluid becomes turbulent. Turbulent flow ischaracterized by chaotic unsteady vortices of manydifferent sizes that increase mixing, friction, andenergy dissipation. In turbulence, the rate of energydissipation increases to approach that of the incom-ing kinetic energy. Turbulent flows are not com-mon in the healthy circulatory system; rather, flow-ing blood more commonly develops jets andswirling motions.
The term vorticity (typically represented by theGreek letter �) is a mathematical term that physi-cally corresponds to (twice) the local angular veloc-ity of a fluid particle. Although a vortex is acompact region characterized by the accumulationof vorticity (Fig. 1B), a shear layer (or a vortex layer)is an elongated layer of vorticity. In general, thespatial distribution of vorticity, as well as its devel-opment and evolution, can facilitate an understand-ing of the key features of blood motion in largevessels and heart chambers. Therefore, most fluiddynamic studies focus on the lifespan of vorticity,from its generation to its organization in the formof vortices, and to its dissipation.
Vorticity does not appear spontaneously in a flowfield; it develops in the form of a shear layeradjacent to the tissue (boundary layer) as a conse-quence of the velocity difference between the flowand the solid boundary. Where streamlines separatefrom the wall, the fluid tends to curl into a vortex ora cascade of vortices (Fig. 1C). Streamline separa-tion and vortex formation are typical of the flowinto the heart cavity from a vein or valve or from anarrow to a more dilated vessel segment. Examplesinclude the flow into the carotid sinus and the flowbeyond a heart valve or vascular stenosis (Fig. 1D).The presence of such formed vortices affects pres-sure distribution and shear stresses and is crucial forrelating blood motion to pathology (3).
An expanded version of this section, along withassumptions used in numerical modeling, has beenincluded in the online Appendix.
Blood Flow Visualization Strategies
The following section provides an overview of thecurrently available flow quantification techniquesfor use in both research and clinical applications.Table 1 lists some of the advantages and disadvan-tages of each method.Flow visualization using CMR. Using the velocity-ncoded CMR phase-contrast technique (4–6),
lood flow velocities can be measured in any direc-imagiDownloaded from
ion without using contrast agents. By using bipolarradients, the signal of the moving blood is in ahase that is proportional to its velocity. Althougheal-time phase-contrast CMR is possible for 2Deasurements, better quality of data is obtained by
ombining the information from several heartbeatssing prospective or retrospective electrocardiogramECG) gating. Retrospective ECG gating withime-resolved, through-plane, and 2D phase-ontrast CMR is mostly used to obtain peak veloc-ties or volumetric flow to determine the strokeolume. Using ECG and respiratory gating, theomplete time-resolved, 3-dimensional (3D), and-directional velocity field can be measured over aolume that covers the complete heart or largeessels (7,8). This 3D cine phase-contrast CMRechnique is popularly called 4-dimensional flowMR. The acquisition time depends on severalarameters: the region of interest, the location inhe body, the spatial and temporal resolu-ion, and the breathing pattern (9,10).urrently, cardiac blood flow can be mea-
ured with a spatial resolution of 3 mm3
and a temporal resolution of 50 ms inapproximately 20 min. Blood flow in theaorta can often be measured at a higherresolution and with a shorter acquisitiontime.
As the complete time-resolved, 3D, and3-directional velocity field is measured, theblood flow can be analyzed using a numberof tools. Three-dimensional streamlines andpathlines (Figs. 2 and 3; Online Video 1)
ave become popular for intuitively visu-lizing time-resolved 3D blood flow. The flowtructures of interest can be automatically charac-erized from the 3D cine phase-contrast CMRsing a pattern-matching approach, for example11). Quantitative volume flow measurements cane obtained by retrospectively extracting the 2Dlanes (12). In laminar flow, the relative pressureeld can be computed from the velocity field in theorta (13) and cardiac chambers (14). Nonlaminarow can be further characterized by the turbulentinetic energy, which is a measure of velocityuctuations (15) that is based on estimating the
ntravoxel SD of the phase-contrast CMR signalagnitude (16).
METHODOLOGICAL RECOMMENDATIONS. CMRis a uniquely versatile yet complex imaging modal-ity. It is not possible to give concise recommenda-tions on the methods of flow acquisition and
A B B
A N D
CMR �
reson
ECG �
Echo-
partic
LV �
LVOT
tract
PIV �
veloci
VFM �
post-processing without relating them to th
by on March 19, 2012 ng.onlinejacc.org
R E V I A T I O N S
A C R O N YM S
cardiac magnetic
ance
electrocardiogram
PIV � echocardiography
le imaging velocimetry
left ventricular
� left ventricular outflow
particle imaging
metry
e re-
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
308
search or clinical questions to be addressed. Thereare trade-offs to be considered among acquisitiontime, spatial resolution, temporal resolution, signal-to-noise ratio, and the comprehensiveness of thedata coverage. Routine breath-hold cine acquisi-tions using steady-state free precession allow somefeatures of the flow to be visualized in 2D planes.Although not real-time, the local variations inthe blood signal can show jet formations, associ-ated para-jet shear, and at least an impression ofsmall-scale flow instabilities within the largerstreams. Importantly, cine imaging also depictsthe myocardial, valve, and vessel wall dynamicsthat are inseparable from the flows they deliverand contain.
The methods and applications of such 4-dimensionalvelocity acquisition (9) are recommended for bothdisplaying the relatively large-scale patterns of mul-tidirectional blood flow that are consistently re-peated from beat to beat as well as relating flows indifferent cardiovascular compartments to one an-other. However, these methods require relativelylong acquisition times of about 8 to 25 min. Eachreconstructed movie loop is not in real-time butrepresents flow that is effectively phase-averagedover hundreds of heart cycles, and, hence, the movie
Figure 1. Basic Principles in Fluid Mechanics
(A) Blood motion in a straight vessel under a pressure gradient balcenter and gradually decreases to zero at the wall due to viscous alarger near the wall. (B) A vortex (left) is a region of compact vorticbetween 2 streams with different velocities. (C) The deceleration ofwith clockwise vorticity (red). The separated vorticity (below) tendsrating vortex reduces the wall shear stress that may even reverse, wthe wall beneath the vortex. (D) Vortex formation occurs when thea sharp expansion.
loop does not record any small-scale instabilities or
byimaging.onlinejacc.orgDownloaded from
other beat-to-beat variations in the flow. The mul-tidirectional velocity data, although impressivelycomprehensive, may need to be supplemented byeither more selective flow imaging at high temporaland spatial resolutions or computational fluid dy-namic simulations to reach conclusions regardingsmall-scale and time-varying flow features. Thissolution may be important to improve the calcula-tion of parameters such as fluid energy dissipation,fluid wall shear stresses, and the partitioning versusmixing of adjacent blood volumes or streams.Flow visualization using echocardiography. Ultra-sound flow visualization techniques, with hightemporal resolution and relatively low cost, makeit suitable for routine clinical use. Recent ad-vances in imaging techniques for mapping flowusing echocardiography techniques are discussedhere.
COLOR DOPPLER ECHOCARDIOGRAPHY. Colorflow was developed in the mid-1980s using multiplerange-gated pulsed Doppler. In clinical practice,however, the technique has been used as a qualita-tive method for demonstrating abnormal flow pat-terns and superimposed turbulence. Color Dopplermeasures only axial velocities along the line of each
s the energy loss due to friction; the fluid velocity is higher in therence. The shear stresses and the vorticity (shaded profile) area shear layer (right) is an elongated distribution of vorticityflow produces (above) a local thickening of the boundary layerroll up and eventually form a vortex in the main flow. The sepa-h is seen as opposite sign, counter-clockwise vorticity (blue) atndary layer formed for the upstream surface detaches because of
ancedheity;thetohicbou
ultrasonic beam. If it were possible to estimate the
on March 19, 2012
ima g.
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
309
radial velocity component, which is directed at rightangles relative to the axial velocity, then combiningthe axial and radial velocities would allow the trueflow vector at each site to be calculated. The firstmethod proposed to achieve this goal was crossed-beam ultrasound (17). This technique uses an arrayof 3 transducers, with 2 acting as transmitters andthe third acting as a receiver. All the transducers arefocused on the same site but are positioned atslightly different insonating angles. This methodhas been proven to work both theoretically and insimple phantoms, but it has not been developed forroutine clinical applications. More recently, theprinciple of measuring 2 velocities at the same siteto derive the vector has been re-investigated forvascular imaging (18).
An alternative method called echo-dynamography, orvector flow mapping (VFM), has been proposed(19). This technique combines measured axial ve-locities with estimated radial velocities based onphysical principles. The major assumption of thefirst version of VFM was that all the flow alongeach radius within an image can be deconstructed
Table 1. Flow-Visualization Technologies
Phase-Encode
Resolution and coverage relative toall 3D of space
Good spatial resolution iin shorter acquisitionunrestricted access
Coverage relative to all 3-directionalcomponents of velocity
All 3 can be acquired, ora plane placed in any
Temporal resolution Typically 20–50 ms, eachcalculated from acquisheart cycles
Scan time Long (5–20 min) for a daall 3D and all 3-directicomponents but seldo
Breath-holding Breath-hold used for shovelocity acquisitions, onavigation for long acfree breathing
Low-velocity accuracy Low velocities are measuaccurate if high velocito be measured
High-velocity accuracy Measurable up to the chbut only where a streawide enough to includ
Applications Flow visualization and mof volume flow througchambers and large ve
Implanted devices Metal stents or valve ringartifact; most pacemak
2D � 2-dimensions; 3D � 3-dimensions; Echo-PIV � echocardiography particle
into a laminar motion and a vortical component
imagiDownloaded from
with a zero mean. Subsequently, the transversevelocity is computed assuming that such a vorticalcomponent must satisfy the continuity equation(mass conservation) from pixel to pixel, along eachradius, and across all the scan lines in the field of thecolor flow image. Although this method relies onnontrivial assumptions, VFM is often able to de-lineate the main features of left ventricular (LV)intracavitary flow (Fig. 4).
An analogous approach (20) uses speckle track-ing to measure the radial velocity of endocardialwall motion around the left ventricle, achieved byapplying the continuity equation in a way suchthat the resulting radial velocity matches theradial velocity at the wall. This technique involvesseparate acquisitions of the grey-scale data for thespeckle tracking of the wall motion and of the colorflow data for estimating the blood velocity. Thistechnique has also been able to discriminate the majorfeatures of vortex flow within the left ventricle duringthe cardiac cycle, but the published images do notshow small, low-velocity vortices.
All of these methods currently ignore through-
RI
Echocardiography
Echo-PIV C
3D or in 2Ds;
Good spatial resolution in 2Dplanes, and in multiple planes ifrequired
High spgood
r 1, throughntation
Both in-plane componentsrepresented but not the through-plane
Only thedirectransmeas
se isover many
High temporal resolution (4–20 ms),allows assessment of velocityfields even during briefisovolumic contraction andrelaxation phases over fewheartbeats
Good te2D (4low i
t coveringleal-time
Both scan time and offline analysiscan be done over few heartbeatsin minutes
Rapid scvisua
nidirectionalaphragmitions during
Imaging relatively easier duringbreath-hold but also possibleduring respiration
Not req
e but lessalso have
Well-visualized May beor aff
VENC limit,r jet core ishole voxels
May be underestimated Well-reslimit
rementll cardiacls
Flow visualization through all cardiac chambers, amay require use of transesophageal echocardio
use localrule out MRI
Flow can be visualized through implanted cardiacpresence of pacemakers and defibrillators
ging velocimetry; MRI � magnetic resonance imaging; VENC � velocity encodin
d M olor Doppler
n alltime
atial resolution in 2D;resolution in 3D
2, oorie
1 componentted to or from theducer is currentlyurable clinically
phaition
mporal resolution in–20 ms), relativelyn 3D
taseonam r
an times, real-timelization
rt, ur diquis
uired
rablties
underestimatedected by noise
osenm oe w
olved within aliasing
easuh asse
ortic flow evaluationgraphy
s caers
devices, and in the
plane flow in the third dimension. Based on
by on March 19, 2012 ng.onlinejacc.org
tqlvrslow
2vtut
CMR � ca
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
310
comparisons with phase-contrast magnetic reso-nance imaging in apical long-axis planes, the errorintroduced by this simplification is on the order of15% for healthy hearts (20). In addition, the accu-racy is necessarily dependent on the accuracy of theDoppler color data. Improving the basic color flowwith an optimum Nyquist limit to reduce lowervelocities from being filtered out and automatingthe process of de-aliasing the color data will behelpful in future development of these methods(21). Novel techniques for the reconstruction of theblood flow using 2D or 3D Doppler imaging arecurrently under investigation.
ECHOCARDIOGRAPHY PARTICLE IMAGING VELOCI-
METRY. Particle image velocimetry (PIV) is a pop-ular technique to characterize the flow field in fluiddynamics laboratories (22,23). When using PIV,the particle-seeded flow is typically illuminatedwith a light sheet generated by a pulsed laser. Theparticle patterns in the field are tracked frame byframe, and the displacement data are converted tovelocity using the small time-frame between theilluminating laser pulses. It is important to under-
CMR-Derived Pathline Visualization of Cardiac Blood Flow
hrough a flow field track the journey of a particle emitted at ae-frame and incorporate velocity information in color throught phases of the cardiac cycle. By plotting the time-resolved path-esulting traces reflect the dynamics of the 3-dimensional bloodthe cardiac cycle. In this figure, the pathlines of the blood flowfrom the left atrium to the aorta (red-yellow) and from the rightthe pulmonary arteries (blue-turquoise) during a cardiac cycle.rdiac magnetic resonance.
stand that the PIV technique does not track indi- t
byimaging.onlinejacc.orgDownloaded from
vidual particles; rather, it tracks the patterns pro-duced by groups of particles. Ultrasound beams areused as the imaging source in echocardiographyPIV (echo-PIV), which enables visualizing bloodflow within an opaque cardiac chamber. The appli-cation of ultrasound-based PIV was first reportedfor imaging kaolin particles in the study ofsediment-laden flow (24). After the introduction ofultrasound-based PIV, or echo-PIV as it is knownin its current form, as a method for capturing digitalB-mode images of contrast agent particles (25), thetechnique has been successfully explored in bothexperimental (26) and clinical (27) settings. Acareful validation of the technique applied to thecarotid flow was recently developed in the labora-tory environment and included a clinical feasibilityanalysis and CMR comparison (28). Echo-PIVprovides a tool to assess blood flow directions andstreamlines (Fig. 5), map principal flow patterns,and define recirculating regions and vortices withreasonable confidence in a visually reproduciblescheme (26).
Many of the limitations of echo-PIV are relatedto the limitations of current ultrasound technology.For example, the time interval between the 2 framesneeds to be short for optimal cross-correlation,which can only be achieved by reducing the sectorangle and depth variables. Current applicationshave mostly been developed for 2D flow trackingand for a sector width size of 45°. Flow can now bemeasured at a temporal resolution as high as 4 msand with an effective spatial resolution of about 4mm. Due to this high temporal resolution, the flowsequences for the phases of the cardiac cycle, in-cluding the brief intervals of isovolumetric contrac-tion and relaxation, have been characterized (On-line Videos 2 and 3). Although the echo-PIVechnique seems to be a promising approach withualitatively meaningful results, it does have severalimitations. An experimental validation study re-ealed that, depending on the acquisition frameate, high velocities can be underestimated, and thepatial resolution of current ultrasound imagingimits accurate imaging of the small-scale featuresf ventricular flow (26,27), which has implicationshen using echo-PIV for diagnostic purposes.
METHODOLOGICAL RECOMMENDATIONS. DuringD echocardiography, acquiring 3- or 5-chamberiews is helpful for visualizing both the filling andhe emptying flow patterns. The width of theltrasound scans, imaging depth, and spatial andemporal settings have to be optimized to achieve
Figure 2.
Pathlines tchosen timsubsequenlines, the rflow overare tracedatrium to
he highest possible frame rate with a scan angle
on March 19, 2012
ftsTefeat(
aae
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
311
large enough to ensure that the entire blood pool isvisualized. Echo-PIV requires a minimum framerate of approximately 60 to 100 frames/s; otherwise,the higher velocities are underestimated (as a simplerule, the frame rate must be higher than the heartrate). Much higher frame rates (�150 frames/s)may be required to visualize transient flow patternsduring the brief isovolumetric contraction period(Online Video 2). The flow pattern is better visu-alized with the tissue harmonic modality and byregulating the mechanical index according to themanufacturer’s recommendations; as a general rule,a single focal zone at the base of the left ventricle ispreferable. The contrast agent must homogeneouslyfill the cavity for clear visualization of the swirlingflow while avoiding both saturation and black areas.To obtain good results, it is preferable to acquirethe loop during the dilution phase that follows thepeak of bubble concentration using the contrastsignal growth and plateau phases to adjust the timegain compensation curve, gain, and focal zone.In vitro experiments and numerical simulations. Care-ul replication of the heart chambers and vascula-ure in vitro has significantly improved our under-tanding of the physics of blood flow (24,29–33).hese models are mainly reconstructed from
ither anatomical castings or images acquiredrom CMR, computed tomography scanning, orchocardiography (29). PIV is commonly used tossess cardiovascular fluid dynamics in vitro, par-icularly for evaluating new imaging modalities
Figure 3. CMR-Derived Short Streamlines
A streamline is a line that is tangent everywhere to the velocity vechave been computed backward and forward from a 3-chamber pla(B) late diastole, and (C) ejection phases of the cardiac cycle. An isodiastole. CMR � cardiac magnetic resonance. See also Online Video
29,34 –36), heart valves (32,37), and ventricular
imagiDownloaded from
ssist devices (38,39). One important consider-tion when using PIV to assess a cardiovascularvent is the temporal resolution. The temporal
field at a given instant in time. In this figure, short streamlinesthe left ventricle of a healthy volunteer during (A) early diastole,
ace is shown in green that outlines a vortex ring in early and late
Figure 4. Flow Velocity Vector Fields Obtained by UsingVector Flow Mapping
Color Doppler data are decomposed into basic and vortex flow componeflow velocity fields are displayed superimposed on the color Doppler imagvelocity vector at a single point is indicated by the red dot to which the y
torne insurf1.
nts, andes. Theellow line
is attached. Image courtesy of Tokuhisa Uejima, Cardiovascular Institute, Tokyo, Japan.
by on March 19, 2012 ng.onlinejacc.org
tvaoateLladttv
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
312
resolution of conventional PIV systems that isappropriate for most fluid dynamics applicationsare about 15 velocity frames/s. These systems arenot appropriate for assessing cardiovascularevents, and due to their poor time resolution,they cannot capture most flow features in aconsistent and reproducible fashion even if theyare time-averaged over multiple cycles. Only lasersystems with repetition rates around 1,000 Hz,together with correspondingly high-resolution,high-speed cameras, are legitimate for quantita-tive analyses of unsteady cardiovascular systemsin vitro.
Numerical simulations attempt to imitate a dy-namic behavior of a system often to predict thesequence of event. Numerical simulations for car-diovascular flow use a combination of computa-tional methods, including medical imaging tech-niques for determining vascular geometry andtechniques for representing the flow domain as alarge number (typically millions) of discrete pointswhen evaluating fluid properties. In principle, it ispossible to simulate any flow by solving the math-ematical laws governing fluid mechanics or by directnumerical simulations. However, it must be stressedthat, despite the enormous progress that has beenmade in computing power, most cardiovascular flowproblems remain challenging. Numerical simula-tions represent a powerful tool that can complementthe clinical analyses of cardiovascular disease. Thesesimulations can be used either when in vivo data arenot measurable or as a predictive tool for therapeu-
Figure 5. Echo-PIV–Derived Streamlines
Streamlines have been color encoded with kinetic energy derivederived instantaneous velocity data obtained from a healthy voltricle during the (A) early diastole, (B) late diastole, and (C) ejec
tic options. d
byimaging.onlinejacc.orgDownloaded from
Clinical Applications and Research Prioritization
The conventional indexes of cardiac performanceusually do not reveal significant changes untilthere is overt dysfunction, which makes theseindexes less effective for the early diagnosis andtreatment of cardiac disease. Flow, instead, isimmediately affected by changes in cardiac func-tion. Therefore, analyzing the blood flow dynam-ics opens up new perspectives for our understand-ing of cardiovascular physiology and for developing veryearly diagnostic tools (Table 2). The clinical utilityof this hypothesis still requires further studies, bothto understand the added value over conventionalapproaches and to develop appropriate flow-basedindexes for applications in various cardiacpathologies.Cardiac function abnormalities. Changes in eitherhe LV shape or contractility will alter the intra-entricular flow patterns. In a normal heart, thesymmetric geometry of the left ventricle and therientation of its valves favor vortex formation andutomatic redirection of blood from the LV inflowo the outflow. This efficient arrangement avoidsxcessive regional wall stress and ensures an optimalV pump performance. In patients with a dilated
eft ventricle, the transmitral flow is often directedlong the free wall, which gives rise to a well-eveloped circular flow pattern that turns towardhe septum and the outflow tract during diastole. Athe end of diastole, blood located close to the leftentricular outflow tract (LVOT) in patients with a
om echocardiography particle imaging velocimetry (echo-PIV)–er. The flow is shown along the long-axis view of the left ven-phases of the cardiac cycle. See also Online Videos 2 and 3.
d fruntetion
ilated left ventricle is directed more perpendicular
on March 19, 2012
tttsimC
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
313
to the LVOT than in the normal left ventricle, inwhich the blood flow is more directed toward theLVOT (40). Blood flow in the left ventricle hasbeen suggested to contain direct flow routes andrecirculating flow patterns, all of which contributeto the efficiency of the pumping function (1,41–44).The presence of a physiological vortex formationprocess supports such flow routes, which avoidsexcess turbulence and energy dissipation. Movingfurther ahead, it can be hypothesized that aproper vortex formation process is associated withthe heart’s ability to adapt to varying conditionswhereas, in its absence, the flow may develop adisordered motion, improper redirection duringcontraction, and nonphysiological peak pressures.The relationship between abnormal vortex for-mation and the progression of LV remodelingand symptoms in patients with heart failureremains to be investigated.
Flow visualization may also be useful in under-standing pathophysiological mechanisms in LV re-modeling. One such mechanism includes, amongothers, the mechanism of outflow tract obstructionin patients with hypertrophic cardiomyopathy. Inthe normal left ventricle, the formation of a trans-mitral vortex helps to maintain the position of themitral leaflets close to the posterior wall and assistsin directing the upcoming systolic stream towardthe outflow tract. In hypertrophic cardiomyopathy,however, the anterior displacement of the papillarymuscles moves the entire mitral apparatus adjacentto the outflow tract, which reverses the direction ofthe transmitral vortex and promotes the anteriormotion of the mitral leaflets during systole due tothe creation of drag forces. This phenomenonrequires further investigation (45,46).
Flow visualization may also have application in
Table 2. Clinical Applications of Flow Visualization Techniques
Clinical Condition
LV systolic function Paths and kinetic energy changes of bdilated/hypertrophic LV remodelingoptimization of resynchronization t
LV diastolic function Transmitral flow patterns and spatial d
Atrial function Flow features for stratifying risks of leincluding Fontan circulation
Valvular diseases Relationship of regurgitation jet on tuon valvular flow direction and LV re
Aorta Relationship of flow characteristics andissection in Marfan syndrome, retraortic reconstruction surgeries
Pulmonary artery Characterization of flow features assoc
LV � left ventricular.
the assessment of cardiac dyssynchrony. Quanti- o
imagiDownloaded from
tatively evaluating cardiac efficiency by usingintraventricular flow dynamic analysis before andafter resynchronization may provide useful indi-cators of clinical severity and the eventual successof therapy. Moreover, the ability of the techniqueto discriminate the beat-to-beat changes in thecardiac vorticity and intracavitary pressure gradi-ents could be used to fine-tune biventriculardevices.
LV flow visualization can provide a novel andpromising method to assess the continuum of sys-tolic and diastolic function. Initiation of LV dias-tole is related to the release of the elastic energystored during the previous systole, and this elasticenergy generates suction and contributes to normalLV filling (47). Early diastolic suction that existsunder physiological conditions may have greaterroles during exercise and tachycardia. Therefore,the causal relationship between LV wall recoil,transmitral flow, and intraventricular pressure gra-dient demonstrates the importance in defining theperformance of the left ventricle during diastole.Quantification of diastolic function and intraven-tricular pressure gradients using echocardiographyis typically based on measuring the changes in theventricular wall, such as the velocity of wall relax-ation (48) and the pulsed wave Doppler pattern ofthe velocity at the mitral tips. There is a need forstudies that relate the well-validated parameters ofdiastolic function (i.e., �, pressure–volume loops,he stress–strain ratio) to the morphological data ofhe resulting flow energetics. The formation ofransmitral vortices is highly sensitive to the pres-ure gradient whether due to drop in LV pressure orncreases in the left atrial pressure and, therefore,
ay be a sensitive marker of diastolic function.onditions that interfere with the normal sequence
Potential Applications
d flowing into the left ventricle for understanding development andsessing stagnant flow and risk of thrombus formation, assessment ofpy and assist devices
ibution of intraventricular pressure gradients, shear stress, and kinetic
rial clot formation, efficiency of flow in congenital heart diseases,
ence and energy dissipation, effects of valve repair and prosthetic redeling
ear stress with risks of aortic atherosclerosis, risks of aortic dilation ande flow from descending aorta and risks of cerebral embolism, optim
d with pulmonary artery remodeling in pulmonary hypertension and
loo progression of, as LV dyssynchrony,hera
istr energy
ft at
rbul placement surgerymo
d sh dogra ization of
iate thrombus formation
f regional contraction and relaxation can be ex-
by on March 19, 2012 ng.onlinejacc.org
sgsatsdsbpacn
vaarbsRtbjttrershid
tsatal
ti
hSpagpvhdtppesotda
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
314
pected to alter early diastolic filling and the vortexformation process. Moreover, a proper character-ization of the intraventricular fluid dynamics, ener-getics, and space-time distributions of pressure andshear stress may allow recognizing early functionalchanges before they provoke irreversible myocardialdysfunction.Assessment of atrial function. Rather than being aimple conduit, the atria have also been shown toenerate consistent blood flow patterns that arepecific to the phase of the cardiac cycle (49). Fromfluid dynamics perspective, the atrium and ven-
ricle seem to be tightly linked, as are diastole andystole (40,44). Changes due to disease result inecreased efficiency and elevated turbulence inten-ity (40,44,50). The function of the right atrium haseen explored in healthy volunteers (1,51) and inatients with Fontan circulation (52). The flowcross the right side of the heart, in general, isomplex, and knowledge of right heart fluid dy-amics remains rather limited.
The valvular diseases. Any change to the mitralalve, whether due to disease or surgery, directlyffects the LV flow pattern. Some of the problemsssociated with mitral valve prostheses that may beelated to blood flow are platelet activation, throm-us formation, hemolysis, increased myocardialtress leading to LV hypertrophy, and remodeling.egurgitant flow in mitral prolapse has been shown
o result in highly disturbed left atrial flow withoth vortices located in proximity to the regurgitant
et as well as elevated values of turbulent intensity inhe left atrium during systole, which are related tohe regurgitant volume (50). Clinical studies areequired to understand the impact of valvular dis-ases and replacement surgery on unfavorable LVemodeling that has been suggested by numericalimulation (53). Such an understanding would en-ance the therapeutic decision-making process and
mprove surgical procedures, particularly in cases ofysfunctional ventricles.Controversies still exist regarding the role of
he vortices that develop in the sinuses of Val-alva in both optimizing aortic valve closure andllowing coronary blood flow. Several studies ofhe downstream flow from prosthetic valves andortic root surgeries have been published in theast 10 years (52,54,55). The impact of changes in
byimaging.onlinejacc.orgDownloaded from
he aortic sinuses and coronary flow remain to benvestigated.Aortic and vascular flow. Aortic blood flow patterns
ave been described in healthy volunteers (9,56).tudies of alterations in aortic blood flow patterns,articularly the vortex and swirl behaviors related toge, coronary artery disease, atherosclerosis, con-enital abnormalities, surgical interventions, andulmonary hypertension, have recently been re-iewed (9). In addition to its enhanced insight intoemodynamics, 3D cine phase-contrast CMR hasemonstrated its advantages in the clinical evalua-ion of aortic coarctation (57). Using 3D cinehase-contrast CMR, retrograde flow from com-lex plaques in the descending aorta can explainmbolisms to all brain territories and the resultingtrokes (58). Larger studies and follow-up imagingf patients with abnormal flow patterns are neededo further evaluate the associations that have beenrawn between altered flow in the thoracic aortand disease outcomes.
Conclusions
The development and validation of novel flowvisualization techniques may enable clinical imag-ing and analysis of multidirectional flow in thecardiovascular system. The identification, verifica-tion, and interpretation of flow-related physiologyin normal and abnormal states may provide addi-tional/incremental insights into a range of cardio-vascular diseases. With advances in imaging, thetime is perhaps ripe for further research into thediagnostic and prognostic impact of intracardiacand vascular flow structure.
AcknowledgmentsThis paper was developed after the internationalmeeting on cardiovascular fluid mechanics theoryand investigative techniques held at University ofCagliari, Sardinia, Italy, June 27 to 29, 2011. Thedata presented, analyzed, and developed at themeeting were compiled as the consensus docu-ment.
Reprint requests and correspondence: Dr. Partho P. Sen-gupta or Dr. Jagat Narula, The Mount Sinai MedicalCenter, One Gustave L. Levy Place, PO Box 1030, NewYork, New York 10029. E-mail: Partho.Sengupta@
valve and prosthesis geometry on intrinsic flow in mountsinai.org or [email protected].
on March 19, 2012
1
1
1
1
1
1
2
2
2
2
2
2
2
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
315
R E F E R E N C E S
1. Kilner PJ, Yang GZ, Wilkes AJ, Mo-hiaddin RH, Firmin DN, YacoubMH. Asymmetric redirection of flowthrough the heart. Nature 2000;404:759–761.
2. Querzoli G, Pedrizzetti G, Kilner P,editors. Proceedings of EuromechColloquium 529 —CardiovascularFluid Mechanics: From TheoreticalAspects to Diagnostic and Therapeu-tic Support, Cagliari, 27–29 June2011. Università degli studi di Cagli-ari 2011 (ISBN 978-8-8900-5580-5).
3. Kheradvar A, Pedrizzetti GS-V. Vor-tex formation in the cardiovascularsystem. London, UK: Springer-Verlag, 2012.
4. Moran P. A flow velocity zeugmato-graphic interlace for NMR imaging inhumans. Magn Reson Imaging 1982;1:197–203.
5. Bryant D, Payne J, Firmin D, Long-more D. Measurement of flow withNMR imaging using a gradient pulseand phase difference technique.J Comput Assist Tomogr 1984;8:588–93.
6. Pelc N, Herfkens R, Shimakawa A,Enzmann D. Phase contrast cinemagnetic resonance imaging. MagnReson Q 1991;7:229–54.
7. Firmin D, Gatehouse P, Konrad J,Yang G, Kilner P, Longmore D. Rapid7-dimensional imaging of pulsatile flow.Comp Cardiol 1993:353–6.
8. Wigström L, Sjöqvist L, Wranne B.Temporally resolved 3D phase-contrast imaging. Magn Reson Med1996;36:800–3.
9. Markl M, Kilner PJ, Ebbers T. Com-prehensive 4D velocity mapping of theheart and great vessels by cardiovascu-lar magnetic resonance. J CardiovascMagn Res 2011;13:7.
10. Ebbers T. Flow imaging: cardiac ap-plications of 3D cine phase-contrastMRI. Curr Cardiovasc Imaging Rep2011;4:127–33.
11. Heiberg E, Ebbers T, Wigstrom L,Karlsson M. Three-dimensional flowcharacterization using vector patternmatching. IEEE Trans Vis ComputGraph 2003;9:313–9.
12. Roes SD, Hammer S, van der GeestRJ, et al. Flow assessment throughfour heart valves simultaneously using3-dimensional 3-directional velocity-encoded magnetic resonance imagingwith retrospective valve tracking inhealthy volunteers and patients withvalvular regurgitation. Invest Radiol2009;44:669–75.
13. Yang GZ, Kilner PJ, Wood NB, Un-derwood SR, Firmin DN. Computa-tion of flow pressure fields from mag-
netic resonance velocity mapping.Magn Reson Med 1996;36:520–6.Dow
4. Ebbers T, Wigstrom L, Bolger A,Wranne B, Karlsson M. Noninvasivemeasurement of time-varying three-dimensional relative pressure fieldswithin the human heart. J BiomechEng 2002;124:288–93.
5. Dyverfeldt P, Kvitting JP, SigfridssonA, Engvall J, Bolger AF, Ebbers T.Assessment of fluctuating velocities indisturbed cardiovascular blood flow: invivo feasibility of generalized phase-contrast MRI. J Magn Reson Imaging2008;28:655–63.
6. Dyverfeldt P, Sigfridsson A, KnutssonH, Ebbers T. A novel MRI frame-work for the quantification of anymoment of arbitrary velocity distribu-tions. Magn Reson Med 2011;65:725–31.
7. Fox KM, Patel RG, Graham GR, etal. Multiple and single ventricular sep-tal defect. A clinical and haemody-namic comparison. Br Heart J 1978;40:141–6.
8. Swillens A, Segers P, Torp H, Lovstak-ken L. Two-dimensional blood velocityestimation with ultrasound: speckletracking versus crossed-beam vectorDoppler based on flow simulations in acarotid bifurcation model. IEEE TransUltrason Ferroelectr Freq Control 2010;57:327–39.
9. Ohtsuki S, Tanaka M. The flowvelocity distribution from the Dopp-ler information on a plane in three-dimensional flow. J Visual 2006;9:69 –72.
0. Garcia D, Del Alamo JC, Tanne D, etal. Two-dimensional intraventricularflow mapping by digital processingconventional color-Doppler echocar-diography images. IEEE Trans MedImaging 2010;29:1701–13.
1. Muth S, Dort S, Sebag IA, Blais MJ,Garcia D. Unsupervised dealiasingand denoising of color-Doppler data.Med Image Anal 2011;15:577–88.
2. Adrian R. Particle-imaging techniquesfor experimental fluid mechanics. AnnRev Fluid Mech 1991:261–304.
3. Willert C, Gharib M. Digital particleimage velocimetry. Exp Fluids 1991:181–93.
4. Crapper M, Bruce T, Gouble C. Flowfield visualization of sediment-ladenflow using ultrasonic imaging. DynamAtmos Oceans 2000:233–45.
5. Kim H, Hertzberg J, Shandas R. De-velopment and validation of echo PIV.Exp Fluids 2004;36:455–62.
6. Sengupta PP, Khandheria BK, KorinekJ, et al. Left ventricular isovolumic flowsequence during sinus and pacedrhythms: new insights from use of high-resolution doppler and ultrasonic digitalparticle imaging velocimetry. J Am Coll
Cardiol 2007;49:899–908.byimaging.onlinejacc.orgnloaded from
27. Hong GR, Pedrizzetti G, Tonti G, etal. Characterization and quantificationof vortex flow in the human left ventri-cle by contrast echocardiography usingvector particle image velocimetry. J AmColl Cardiol Img 2008;1:705–17.
28. Zhang F, Lanning C, Mazzaro L, et al.In vitro and preliminary in vivo valida-tion of echo particle image velocimetryin carotid vascular imaging. UltrasoundMed Biol 2011;37:450–64.
29. Kheradvar A, Houle H, Pedrizzetti G,et al. Echocardiographic particle im-age velocimetry: a novel technique forquantification of left ventricular bloodvorticity pattern. J Am Soc Echocar-diogr 2010;23:86–94.
30. Berry JL, Santamarina A, Moore JEJr., Roychowdhury S, Routh WD.Experimental and computational flowevaluation of coronary stents. AnnBiomed Eng 2000;28:386–98.
31. Kheradvar A, Gharib M. On mitralvalve dynamics and its connection toearly diastolic flow. Ann Biomed Eng2009;37:1–13.
32. Kheradvar A, Kasalko J, Johnson D,Gharib M. An in vitro study of chang-ing profile heights in mitral biopros-theses and their influence on flow.ASAIO J 2006;52:34–38.
33. Kheradvar A, Milano M, Gharib M.Correlation between vortex ring for-mation and mitral annulus dynamicsduring ventricular rapid filling.ASAIO J 2007;53:8–16.
34. Westerdale J, Belohlavek M, McMa-hon EM, Jiamsripong P, Heys JJ,Milano M. Flow velocity vector fieldsby ultrasound particle imaging veloci-metry: in vitro comparison with opti-cal flow velocimetry. J UltrasoundMed 2011;30:187–95.
35. Firstenberg, MS, Smedira NG, Green-berg NL, et al. Relationship betweenearly diastolic intraventricular pressuregradients, an index of elastic recoil,and improvements in systolic and di-astolic function. Circulation 2001;104Suppl 1:I330–5.
36. Kitajima HD, Sundareswaran KS, Teis-seyre TZ, et al. Comparison of particleimage velocimetry and phase contrastMRI in a patient-specific extracardiactotal cavopulmonary connection. J Bio-mech Eng 2008;130:041004.
37. Stühle S, Wendt D, Houl G, et al.In-vitro investigation of the hemody-namics of the Edwards Sapien trans-catheter heart valve. J Heart Valve Dis2011;20:53–63.
38. Laumen M, Kaufmann T, Timms D,et al. Flow analysis of ventricular assistdevice inflow and outflow cannula po-sitioning using a naturally shaped ven-tricle and aortic branch. Artif Organs
2010;34:798–806.on March 19, 2012
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 5 , N O . 3 , 2 0 1 2
M A R C H 2 0 1 2 : 3 0 5 – 1 6
Sengupta et al.
Cardiovascular Flow Visualization
316
39. Roszelle BN, Deutsch S, ManningKB. Flow visualization of three-dimensionality inside the 12 cc PennState pulsatile pediatric ventricular as-sist device. Ann Biomed Eng 2010;38:439–55.
0. Carlhall CJ, Bolger A. Passingstrange: flow in the failing ventricle.Circ Heart Fail 2010;3:326–31.
1. Kim WY, Walker PG, Pedersen EM,et al. Left ventricular blood flow pat-terns in normal subjects: a quantitativeanalysis by three-dimensional mag-netic resonance velocity mapping.J Am Coll Cardiol 1995;26:224–38.
2. Bolger AF, Heiberg E, Karlsson M, etal. Transit of blood flow through thehuman left ventricle mapped by cardio-vascular magnetic resonance. J Cardio-vasc Magn Reson 2007;9:741–7.
3. Eriksson J, Carlhall CJ, Dyverfeldt P,Engvall J, Bolger AF, Ebbers T.Semi-automatic quantification of 4Dleft ventricular blood flow. J Cardio-vasc Magn Reson 2010;12:9.
4. Eriksson J, Dyverfeldt P, Engvall J,Bolger AF, Ebbers T, Carlhall CJ.Quantification of presystolic bloodflow organization and energetics inthe human left ventricle. Am J Physiol2011;300:H2135–41.
5. Lefebvre XP, He S, Levine RA, Yo-ganathan AP. Systolic anterior motionof the mitral valve in hypertrophiccardiomyopathy: an in vitro pulsatileflow study. J Heart Valve Dis 1995;4:422–38.
6. Lefebvre XP, Yoganathan AP, LevineRA. Insights from in-vitro flow visu-
alization into the mechanism of sys-imagDownloaded from
tolic anterior motion of the mitralvalve in hypertrophic cardiomyopathyunder steady flow conditions. J Bio-mech Eng 1992;114:406–13.
7. Courtois M, Kovacs SJ, LudbrookPA. Physiological early diastolic intra-ventricular pressure gradient is lostduring acute myocardial ischemia.Circulation 1990;81:1688–96.
8. Yellin EL, Meisner JS. Physiology ofdiastolic function and transmitralpressure-flow relations. Cardiol Clin2000;18:411–33, vii.
9. Fyrenius A, Wigstrom L, Ebbers T,Karlsson M, Engvall J, Bolger AF.Three dimensional flow in the humanleft atrium. Heart 2001;86:448–55.
0. Dyverfeldt P, Kvitting JP, Carlhall CJ,et al. Hemodynamic aspects of mitralregurgitation assessed by generalizedphase-contrast MRI. J Magn ResonImaging 2011;33:582–8.
1. Fredriksson AG, Zajac J, Eriksson J,et al. 4-D blood flow in the humanright ventricle. Am J Physiol HeartCirc Physiol 2011;301:H2344–50.
2. Markl M, Geiger J, Kilner PJ, et al.Time-resolved three-dimensionalmagnetic resonance velocity mappingof cardiovascular flow paths in volun-teers and patients with Fontan circu-lation. Eur J Cardiothorac Surg 2011;39:206–12.
3. Pedrizzetti G, Domenichini F, TontiG. On the left ventricular vortex re-versal after mitral valve replacement.Ann Biomed Eng 2010;38:769–73.
4. Lim WL, Chew YT, Chew TC, LowHT. Pulsatile flow studies of a porcine
bioprosthetic aortic valve in vitro: PIVby on March 19, 2012ing.onlinejacc.org
measurements and shear-inducedblood damage. J Biomech 2001;34:1417–27.
5. Brücker C, Steinseifer U, SchröderW, Reul H. Unsteady flow through anew mechanical heart valve prosthesisanalysed by digital particle image ve-locimetry. Meas Sci Technol 2002;13:1043–9.
6. Kilner PJ, Yang GZ, Mohiaddin RH,Firmin DN, Longmore DB. Helicaland retrograde secondary flow pat-terns in the aortic arch studied bythree-directional magnetic resonancevelocity mapping. Circulation 1993;88:2235–47.
7. Hope MD, Meadows AK, HopeTA, et al. Clinical evaluation ofaortic coarctation with 4D flow MRimaging. J Magn Reson Imaging2010;31:711–18.
8. Harloff A, Simon J, Brendecke S, etal. Complex plaques in the proximaldescending aorta: an underestimatedembolic source of stroke. Stroke2010;41:1145–50.
Key Words: contrast y dopplery echocardiography y flow ymagnetic resonance y vortex.
‹ A P P E N D I X
For supplementary information on this topic
(including assumptions used in numerical mod-
eling), and supplementary videos and their
legends, please see the online version of this
article.
To participate in this CME activity by taking the quizand claiming your CME credit certificate, please go to
www.imaging.onlinejacc.organd select the CME tab on the top navigation bar.
Online Appendix for the following JACC: Cardiovascular Imaging article
TITLE: Emerging Trends in Cardiovascular Flow Visualization
AUTHORS: Partho P. Sengupta, MD, DM, Gianni Pedrizzetti, PHD, Philip J. Kilner, MD, Arash
Kheradvar, MD, PHD, Tino Ebbers, PHD, Giovanni Tonti, MD, Alan G. Fraser, Jagat Narula, MD,
PHD
___________________________________________________________________________
APPENDIX
Basic Concepts (Expanded online version)
The distinctive property of fluids is the development of internal stresses in response to a relative
velocity between nearby elements. This behavior is due to a property of fluids called viscosity
causing a friction during the sliding between individual fluid elements. Given this frictional
phenomenon, the mechanical energy used to deform the fluid elements is lost for dissipation and
transformed into heat. In virtue of this, the natural way of fluid flowing is a balance between the
energy gradient pushing the fluid from upstream to downstream and the energy dissipated during
such motion. Like a river that has a higher energy uphill that is dissipated through friction at its
bed, blood presents a pressure drop along the vessel that is balanced with wall shear stresses.
Theoretically, the fluid moves along parallel laminas whose velocity is larger values at the center
of the vessel and decreases for friction approaching the boundary.
The laminar type of flow is, however, not common. First, the laminar flow is unstable when its
kinetic energy (proportional to the square of its velocity V) exceeds its ability to dissipate energy
(roughly proportional to νV/D; where D is the diameter and ν is the kinematic viscosity that for
blood is about 0.033 cm2/s). Thus when this ratio, known as the Reynolds number Re=VD/ν,
becomes larger than a critical value (that is around 2500 in steady flow in a circular duct) the
laminar motion is unstable and the fluid becomes turbulent. The flow develops a swirling
behavior that increases friction in order to reach a balance with the large injected energy.
Turbulent flow are also not common in the circulatory system, a weak turbulence briefly appears
just in the ascending Aorta, and sometime in large hearts, prevalently during the deceleration
phases when friction needs to increase. More frequently, the flow develops vortices, jets, and
swirling motion that, when present, have a fundamental impact on the fluid dynamics and its
interaction with the surrounding tissues. In order properly evaluate such vortical flows it is useful
to introduce a few fundamental driving concepts.
Vorticity is the key quantity in fluid dynamics; velocity alone is not able to evidence the
underlying dynamical structure of a flow field, made of stresses, mixing, or vortices. Vorticity
(typically indicated with the Greek letter ω) physically corresponds to (twice) the local angular
velocity of a fluid particle but its spatial distribution is the actual skeleton of a flow field. A
vortex, which is loosely described as a fluid structure that possesses circular or swirling motion,
is more properly defined as compact region where vorticity has accumulated. A shear layer, that
separates streams with different velocity, is actually a layer of vorticity, a vortex layer (Figure 1,
panel B). Vortices and vortex layers are the fundamental structures in flow fields. Their different
three dimensional arrangements and combinations give rise to the complexities of all evolving
flows.
Vorticity, however, cannot appear spontaneously within a flow; it develops at the wall only, in
the form of a vortex layer due to the velocity difference between the outer flow and the fluid
attached to the wall for viscous adherence. Such a boundary layer has a central role in fluid
mechanics as it represents the unique possible source of vorticity and its detachment from the
wall is the unique mechanism allowing vorticity to enter into the flow. The phenomenon of
boundary layer separation occurs in consequence of a local deceleration of the flow, as it may be
due to a vessel enlargement, a sharp turn or, in general, an adverse pressure gradient. When
velocity decreases downstream, the boundary layer thickness tends to increase because of
incompressibility. This emerging vorticity is strained by the higher velocity and enters into the
flow as a tongue of vorticity that tends to roll-up into a compact vortex structure and eventually
detaches from the original boundary layer. The Reynolds number, again, reflects the ratio of
flow strength with viscous smoothing, therefore the higher Re the stronger and sharper the
vortices are.
This vortex formation process is the natural phenomenon occurring from cardiac valves, or in
vessels presenting significant flow decelerations, like sometime in the carotid sinus or in
branching regions. Pathologic vortex formation develops from stenosis, when these are sharp
enough, in aneurisms, or in consequence of surgeries affecting the flow. The process of boundary
layer separation modifies the vorticity at the wall and therefore the distribution of wall shear
stress that correlates with atherosclerosis. Formed vortices alter the distribution of pressure on
the surrounding tissues and energy dissipation; they can support or adverse to the proper
redirection of flow paths. In any case, vortices profoundly influence the dynamics of the fluid
and any flow realization involving vortices demands a careful analysis of their presence and
impact.
Numerical Modeling
Numerical models are grounded on the assumption that a clinical problem can be expressed in
mathematical terms; afterwards numerical techniques are invoked as solution methods. The
mathematical formulation typically assumes that blood is a continuous material, an assumption
that fails in small vessels or when exploring phenomena occurring at small scales where the
corpuscular nature of blood cannot be neglected. Numerical approaches involving fluid-tissue
interaction also require solving the equations for the surrounding soft tissues, like myocardium or
valvular leaflets. Tissue mechanics introduces a problematic issue because the mechanical
properties of the tissue are not highly variable, anisotropic and inhomogeneous, and they are not
accessible for measurements in vivo. Finally, every solution of spatially extended problems
(involving partial differential equations) crucially depends on the boundary conditions that, in
turn, require a careful definition of the boundaries themselves. The accuracy of actual boundaries
in numerical simulation are limited by the accuracy of imaging methods and therefore cannot
include details (like taberculae, or geometrical details of cardiac valves) whose relevance must
be preliminarily evaluated.