Article Text

Download PDFPDF

Congenital heart disease
Three dimensional echocardiography in congenital heart disease
  1. Joseph John Vettukattil
  1. Correspondence to Dr Joseph John Vettukattil, Paediatric Cardiology, Wessex Cardiac Centre, Southampton University Hospital NHS Trust, Southampton, SO16 6YD, UK; joseph.vettukattil{at}

Statistics from

Ultrasound imaging of the human heart has undergone revolutionary changes along with recent strides in computing power. Since the wider acceptance of two dimensional (2D) echocardiography in the 1970s, progress in this field had slowed to some extent. However, the quest for three dimensional (3D) ultrasound imaging of the heart began in the early 1960s when Baum and Greenwood introduced the concept by imaging the orbit using a series of parallel slices.w1 It was not until 1974, when Dekker and colleagues sought to construct a 3D model of the heart using a mechanical spatial locator,1 that the concept became more realistic. Their model was limited to an open chest with fixed point imaging, requiring all the desired images to be obtained from one location—an extremely slow and primitive process suitable only for research.

In 1986 Martin and colleaguesw2 used a micromanipulator controlled transoesophageal transducer which marked the beginning of 3D transoesophageal echocardiography (3DTOE). In 1991, Kuroda et al2 described a 3D system that rotated the TOE probe, and simultaneously Woolschlager et alw3 described a TOE system that was able to take serial slices. Further development of a rotating array like a propeller or a fan, parallel to the imaging plane, overcame the problem of the small ultrasound window. In 1989, Raqueno reconstructed the conventional 2D colour flow Doppler images into 3D volumes. TomTec (Unterschleissheim, Germany) converted colour velocity data through a post-processor to assign different colours and used a transparency slider to give the appearance of ‘see through’ jets.

In 1990, Von Ramm and Smithw4 from Duke University used a real-time volumetric 3D system with a matrix array probe. This utilised parallel processing to obtain pyramidal volume which displayed multiple image planes. In this model 2D arrays steered the sound over an entire pyramidal volume, allowing electronic steering and focusing in both elevation and azimuth. Initially, 512 elements were used, 256 for transmission and 256 for receiving. Different images acquired had to be aligned using mathematical interpolation and the gaps were filled. Currently used matrix array probes have >3000 active elements to produce true real-time 3D echo images, while the most recent probes can provide real-time 3D colour.

Early ultrasound systems permitted 3D acquisition, but manipulation of the data required further developments. In the early 1990s TomTec developed a commercially available offline analysis system that could accept datasets from different vendors. Later, online image manipulation was available on the Philips 7500 system followed by the IE33 system with advanced calculations through Qlab (Philips Medical Systems, Andover, Massachusetts, USA). Currently, Siemens, GE Medical, and Toshiba have emerged with comparable alternative systems.3 This discussion is mainly based on the IE33 system.

Clinical application of 3D imaging

Real-time 3D echocardiography (RT3DE) is sometimes referred to as 4D, when the dimension of time is taken into consideration. It is a unique method of accurately visualising the dynamic morphology of the heart. Not only does it display moving images in 3D, but it incorporates the biometric datasets frozen in time, like iris photo-graphy or finger printing. This enables the cardiologist to bring the frozen virtual heart to life and to dissect it, time and time again without corrupting or altering the preserved information. This helps to compare pre- and postoperative anatomy and enhance learning by direct correlation with intraoperative findings. It is also possible to share the datasets electronically between professionals at different geographical locations where data can be independently analysed without the need for transferring patients.

RT3DE has revolutionised the clinical management of congenital heart defects. The technique provides additional information that substantially alters clinical management in many patients.4 5 Even though most currently available echocardiography systems come with the potential for 3D imaging, its clinical utilisation is limited to a few centres which have developed the expertise to implement it for surgical or transcatheter interventions.

Image acquisition

Acquisition of 3D images may be from transthoracic, transoesophageal, or epicardial surfaces. While most interventions for structural heart disease require advanced planning, transthoracic images ought to be acquired as part of 2D imaging in the outpatient setting. Other settings in which images may be acquired include ventilated patients (preoperative or intensive care). Epicardial images are acquired from the open chest perioperatively.

After selection of an appropriate 3D probe, as in 2D imaging, the received images are displayed on the screen in 2D mode. Depending on the clinical need, either live 3D, multiplane, full volume 3D loops (FVL), or colour 3D mode is selected and the corresponding images acquired. Depending on clinical necessity and urgency, further image manipulation is performed.

Practical points for 3D image acquisition

Probe position

The best window for 3D image acquisition is the location from where the best possible image of the structure under evaluation is obtained. Ideally, the ultrasound beam should align perpendicular to the structure under investigation. For example, to acquire the structural details of the mitral valve and sub-valve apparatus, the probe is placed at the position of the apical impulse with the patient in the lateral position. If the surgical view of the mitral valve from the left atrium is required, the best image would be obtained from the parasternal position centreing on the mitral valve.

Gain setting

The gain is usually set high, aiming for uniform echogenicity of the structure under evaluation, and the controls are adjusted to get the best blood tissue separation. Other important aspects of acquisition are: centreing and choosing appropriate elevation, and full visualisation of the structure of importance in two orthogonal planes. All images should avoid movement artefacts and ideally be synchronised with the ECG and respiration. Always ensure that a few FVLs in both colour and grey scale of the anatomic structure under evaluation are obtained for post-processing. A good 2D image is a precursor for good 3D. Live 3DE for interpretation of structural heart disease can often be misleading and is best avoided, except in transoesophageal echocardiography (3DTOE) with zoom mode.


The resolution of 3DTOE images is far superior to transthoracic 2D images, and better anatomic delineation is now possible with the high spatial and temporal resolution. However, patient size can limit the use of 3DTOE, as it is currently recommended only for those over 25 kg, although images have been successfully obtained in complex cases in children weighing about 20 kg. Practical use of 3DTOE is mainly for the assessment of complex defects where the surface 3D resolution is suboptimal or for interventional procedures. These include closure of atrial or ventricular septal defects,w5 transcatheter aortic valve implantation,w6 trans-septal interventions like paravalvular leak,w7 mitral valve annuloplasty,6 or left atrial appendage occlusion.w8 Real-time 3D zoom mode has significantly enhanced the capability of RT3DE by visualising live anatomic details for transcatheter interventions and defining details of the cardiac pathology.

Colour 3D

The availability of live colour 3D imaging has further enhanced the clinical application of 3D imaging. This includes quantification of regurgitant lesions, defect sizing, colour 3D angiography,7 8 and for the differentiation of artefacts from anatomic defects. Direct comparison of 3D echocardiography (3DE) and colour 3D images, acquired from identical positions and planes, helps to delineate gain related dropouts from actual defects.


Image manipulation on the system

RT3DE can be a single volume dataset with low resolution or a compilation of ECG gated 3D datasets stitched together to display a full volume dataset from four to seven consecutive cardiac cycles. On live 3D imaging, on zoom mode, a smaller section of the live 3D window can be viewed in detail. Most recent developments in 3D imaging include live colour 3D and multiplane cropping.

Post-processing is also possible on most currently available 3D equipment. For example, the Philips IE33 system uses built-in Qlab software for instant image manipulation. Using this mode, acquired datasets can be further dissected to analyse the anatomy of the structure of interest or to quantify the lesion. This facility allows defect sizing during 3D assessment of intracardiac defects like atrial and ventricular septal defects or paravalvular leaks during transcatheter closure. The details are elaborated under offline analysis.

Offline analysis

Offline analysis is software dependent. The most commonly used software is Qlab and Image Arena (TomTec). Two different techniques are used for the evaluation of the 3D morphology during post-processing: the fixed plane approach, and multi-planar reformatting (MPR).

Fixed plane approach (cropping box)

In this technique, the 3D dataset is displayed in a cube (pyramid in cube) and the sides of the cube move in a fixed plane, cutting the pyramidal 3D dataset of the heart from all six aspects. There is also provision of a free moving crop adjustment plane which could be used for further image dissection. Though the fixed plane approach is easier to perform, this is not ideal for clinical use as it is fraught with significant errors of interpretation. Since the cardiac structures are not cut in anatomic planes, they may be inappropriately cut or lost, resulting in data that are often misleading or inaccurate. Superimposition of artefacts or structures beyond the plane may give the impression of cardiac anomalies. The best way to analyse the dataset and reconstruct 3D images from it is by using MPR.4 9 All discussions in this article will be based on this technique.


The most important aspect of 3DE is its ability to slice the dynamic cardiac structures in infinite planes through the three dimensions. This method of analysing the anatomy is termed as ‘multi-planar reformatting’ or ‘multi-plane review’. We have improvised this technique of moving the slicing planes simultaneously in an anatomically appropriate manner throughout the cardiac cycle, displaying the images attitudinally appropriate for visualising the structure under evaluation. This technique is most useful in studying and understanding cardiac morphology, especially when the resolution of the images are poor and a visually useful image may not be obtainable for 3D display. MPR may be considered equivalent to anatomic dissection of a pathology specimen with its ability to preserve the specimen despite repeated slicing. It also helps as a mode of transition through multiple frames of familiar 2D images to 3D images, having the advantage of an added plane displaying the depth aspect. When image resolution is poor, especially with transthoracic images, MPR helps to differentiate true anatomic structures from artefacts. Apart from delineation of structural anatomy, the other important application of MPR is in defect sizing and quantification of regurgitant lesions or paravalvular leaks.

Technique of MPR

The purpose of MPR is to interpret and reconstruct cardiac morphology accurately for display while preserving the anatomic planes of dissection and orientation. There are three important steps involved in using MPR: alignment, analysis, and 3D display.


Using post-processing software, the stored FVL is brought to display on screen. The three dissecting planes are adjusted, focusing on the structure of anatomic interest frozen in the phase of the cardiac cycle which displays its details best. For example, if the mitral valve is being evaluated to assess the degree of prolapse, then an end systolic frame is taken from a dataset acquired from the left atrial (LA) view, whereas to study the supra-mitral membrane, a diastolic frame viewed from the LA is desirable. Once the frame is chosen, one of the dissecting planes is brought to the centre of the structure under evaluation to cut it along its long axis (sagittal plane). Another cutting plane is then brought perpendicular to this plane, cutting the structure along its long axis (coronal plane). The third plane is then brought to transect both the above planes at their short axis. Each plane of dissection is continuously readjusted to obtain best visualisation of the anatomy.


Moving one plane of dissection reformats the cardiac structures dissected by that plane into the corresponding position and displays it in a panel representing that plane. Anatomic variation brought about by this change is carefully observed. This action is repeated by moving each plane until structural details are clearly understood. Sometimes different volumes may be analysed to confirm that the observation is not due to artefacts. If the same structural differences are seen in all the corresponding planes in more than one dataset, then that lesion is considered real. The dynamic details of the lesion are studied further by unfreezing the structure and carefully observing it throughout the cardiac cycle. Once adequate knowledge about the lesion is obtained, it is further interpreted on the basis of clinical and haemodynamic data.

3D visualisation

Once the anatomical details and clinical pathology are understood by MPR, 3D reconstruction is performed based on available software. If resolution of the images are not adequate for 3D visualisation, then the MPR images may be displayed as such.

Depending on the software, various specialised products dealing with specific clinical or functional aspects are possible. This includes right ventricular (RV) and left ventricular (LV) 3D volume analysis with semi-automated stroke volume, cardiac output, and dyssynchrony assessment (figure 1). Other applications are calculation of the chamber area, myocardial mass, 3D speckle tracking, mitral valve planimetry, and quantification of annular displacement.

Figure 1

Clinical applications of 3D echocardiography. (A) Dyssynchrony assessment of the left ventricle in a patient with cardiomyopathy. The left ventricular wall in this illustration is divided into 15 segments and the time to peak contraction is plotted. See the variation in the point of peak contraction. The dynamic image illustrates visually the degree of dyssynchrony which is also seen on contraction front mapping. (B) Measurement of left ventricular volume and function using Qlab. Semi-automated tracing of the endocardium in two orthogonal planes and transverse plane is illustrated. (C) Illustration of a common atrioventricular valve with two separate orifices. The zone of apposition between the superior and inferior bridging leaflets can be clearly seen with a coaptation failure at the centre. (D) 3D colour angiography of the great arteries in a patient with right pulmonary artery from the main pulmonary artery and left pulmonary artery from the aorta with a patent arterial duct connecting the pulmonary artery and aorta.

Clinical application of RT3DE in specific congenital heart defects

A detailed discussion of clinical application of 3D echocardiography is beyond the scope of this article. A brief discussion follows.

Reconstruction of cardiac morphology

Accurate reconstruction of the morphology of the semilunar valves or the mitral and tricuspid valves with details of the sub-valve apparatus is possible with 3DE. Visualisation of the dynamic morphology and structural details is unparallelled as it is not possible even during surgery when the heart is stopped. Recently, 3DE has given much insight into the understanding of Ebstein's malformation,5 10 leading to better clinical and surgical management (figure 2). Another important contribution of 3D is in the understanding of complex heart defects, especially the atrioventricular septal defects (AVSDs).11 12 The application of 3D MPR to dissect cardiac anomalies in anatomically appropriate planes has led to biventricular repair of some complex single ventricles.4 This group includes those with ‘unbalanced AVSD’, criss-cross heart, straddling atrioventricular (AV) valves, and double inlet or double outlet ventricles (figure 3). Other clinical applications include visualisation of sub-aortic pathology and complex left ventricular outflow tract obstructions,9 double chambered ventricles, and variation in septal morphology (figure 4) and its defects.13 w9 w10 3DE is also used in understanding the morphological details of the aortic valve and accurate measurement of the valve area, aiding in appropriate catheter interventions14 (figure 5). Mitral valve planimetry and quantification of regurgitation using advanced 3D software is now possible as it “provides complementary information as to the mechanisms and sites of AV valve failure in congenital heart disease”.15 w11 w12 Accurate visualisation of mitral and tricuspid valve defects leads to the assessment of the double orifice mitral valve, paravalvular leak, and other mitral valve pathologies (figure 6). The aortic arch and the great vessels can also be better visualised in children using RT3DE, aiding in the diagnosis of vascular abnormalities including the double aortic arch.w13

Figure 2

Anomalies of the tricuspid valve. (A) 3D echocardiography of a patient with Ebstein's anomaly. Atrialised right ventricle with bifoliate closure of the tricuspid valve, resulting from fusion of the mural and septal leaflets. (B) Rotational anomaly of the tricuspid valve in Ebstein's anomaly illustrated by the short axis view of the tricuspid valve when the mitral valve is seen opening in the long axis of the heart. (C) Large anterosuperior leaflet (ASL) in Ebstein's anomaly illustrating the ‘keyhole’ defect at the ventricular septum due to poor coaptation between the ASL and non-delaminated septal leaflet. (D) Dysplastic tricuspid valve with large right atrium and fenestrated ASL. There is a large deficiency at the coaptation point due to non-delaminated septal leaflet thickened and rolled edges of the dysplastic ASL. There is no evidence of rotational abnormality.

Figure 3

(A) Multi-planar reformatting (MPR) of the subaortic area in a newborn weighing 2.6 kg with interrupted aortic arch and severe sub-aortic stenosis. On 2D cross-sectional imaging the sub-aortic area (arrow) measured 1.7 mm, suggesting severe narrowing. On 3D MPR the orthogonal plane (arrow) shows a square orifice which measured 1.7×3.7 mm, suggesting no significant obstruction on the cut plane. The patient underwent successful biventricular repair without significant residual outflow tract obstruction. (B) MPR of an atrioventricular septal defect showing an abnormal angle formed by the left atrioventricular valve at the crux, suggestive of poor outcome.

Figure 4

3D echocardiography illustration of the varation in the morphology of atrial septal defects. (A) Oval fossa viewed from the right atrium showing a small patent foramen ovale. (B) Aneurysm of the atrial septum protruding into the right atrium with multiple sieve-like defects. (C) Two atrial septal defects with guide wires going through before device closure. (D) Spiral margins of a small secundum atrial septal defect which has adequate margins separating it from adjoining structures but unsuitable for device closure. (E and F) What appears to be a separate foramen ovale and a fenestrated atrial septal defect coalesce together to form a single large defect with a flap on 3D dynamic imaging. Accurate differentiation of these anatomic variations is crucial in successful intervention without complications.

Figure 5

3D imaging of the aortic valve. (A) Trifoliate closure of symmetrical leaflets. (B) Bicuspid aortic valve with a very small right coronary cusp with no evidence of restriction to opening on dynamic imaging. (C) Symmetrical cusps in a bicuspid aortic valve with restricted opening on dynamic imaging. Accurate measurement of the aortic ‘annulus’ is possible with multi-planar reformatting which is significantly different from 2D measurements, especially in deciding the balloon size for valvuloplasty.

Figure 6

Abnormalities of the left atrioventricular valve. (A) 3D visualisation of the morphology of double orifice valve in association with atrioventricular septal defect. (B) Rheumatic mitral valve stenosis with thickened and rolled up valve edges and shortened cordi. Large left atrium demonstrating chronicity of the lesion. (C) Trifoliate left atrioventricular valve demonstrating the zone of apposition between the superior and inferior bridging leaflets with a separate coaptation point differentiating the left atrioventricular valve in an atrioventricular septal defect from a cleft mitral valve. (D) Crescentic appearance of a supra-mitral membrane sparing the aortic aspect.

Volumetry, cardiac deformation imaging, and dyssynchrony assessment

Accurate quantification of volume changes during the cardiac cycle is essential for understanding cardiac physiology and its alteration in disease. Currently, most volumetric methods use geometric assumptions. Accurate volumetry of the left ventricle by RT3DE is now possible16 and it has challenged the need for cardiac MRI, especially in children who may need general anaesthesia. Assessment of RV function is equally important in congenital heart disease, especially when the right ventricle is a systemic ventricle as in the Mustard or Senning operations, congenitally corrected transposition, or in single ventricles with RV morphology. It is also important in assessing ventricular interaction and RV function in tetralogy of Fallot. It is difficult to capture the dilated right heart within the sector width of currently available transducers. However, some authors maintain that RT3D3 “is a very sensitive tool to identify RV dysfunction in patients with congenital heart disease and could be applied clinically to rule out RV dysfunction or to indicate additional quantitative analysis of RV function”.w14

The utility of RT3DE to track volume changes during the cardiac cycle has led to its use in cardiac dyssynchrony assessment and resynchronisation therapy (CRT): “The dyssynchrony measurements by tissue Doppler and RT3DE are not comparable and are unable to predict the response to CRT”.w15 However, quantification of LV mechanical dyssynchrony by 3DE is reproducible and is “an excellent predictor of response to CRT in selected patient cohorts and may be valuable in identifying a target population for CRT irrespective of QRS morphology and duration”.17 Children with LV dysfunction demonstrate increased intraventricular LV dyssynchrony, “in a pattern that is negatively correlated with LV systolic function”.w16 18 Most studies on cardiac dyssynchrony use ECG gated images stitched from multiple cardiac cycles and have limited clinical application due to variation in cardiac cycle length resulting from sinus arrhythmia. Recent developments in cardiac deformation imaging using 3D volumes from a single cardiac cycle and advanced quantification software will significantly improve CRT in congenital heart disease.

Limitations of 3D

3D relies on gain settings and depth delineation to aid 3D visualisation. As gain settings can be confounded with echogenicity of structures, positive or negative artefacts need to be accurately identified. Even though the digital data is in 3D, stereoscopic display is not currently possible. Post-processing is software dependent and online utilisation of the software in standard echocardiography equipment has restricted use. The available software is not focused on the need of the congenital cardiologist as the industry continues to have an adult bias. Frame rate remains low leading to poor optical resolution, though 3DTOE has much improved image resolution. The post-processing technique varies widely among cardiologists and there is an urgent need for a unified protocol. As image analysis is the backbone of RT3DE, training in post-processing is needed for wider acceptance and clinical utilisation.

Future directions

Though some progress has been made in 3DE,14 19 age, sex and body weight matched standardised normal values for cardiac chambers and vessels is an important requirement. Similarly, quantification of myocardial deformation in both biventricular and univentricular circulation in pre- and postoperative states is essential to define and understand the progression of pathology. Improvement in 3D technology may allow 3D display and visualisation using glasses, 3D screen or by holographic projection. The appropriate mode for publication in this field will remain in the electronic media (eg, rather than printed media as the latter cannot display the dynamic and time related changes. Post-processing of 3D datasets on the echocardiography system rather than on a stand alone computer would enable instantaneous definition of detailed morphology and measurements. Ultrasound tracking of the catheter tip would enhance the application of RT3DE in complex interventions. Simplification of post-processing to reconstruct 3D images from post-processed MPR planes would substantially improve its clinical application and would help to move away from the erroneous use of fixed plane cropping.

Characteristics of Ebstein's anomaly demonstrable by 3D echocardiography

  • Rotational anomaly of the tricuspid valve

  • Apical displacement of septal and mural leaflets leading to atrialisation of the right ventricle

  • Failure of delamination of tricuspid valve leaflets

  • Valve dysplasia

  • Abnormalities of tension apparatus

  • Myocardial abnormalities

  • Reduced size of the functional right ventricle

  • Variation in anatomy and function of the systemic ventricle

  • Coexistent cardiac anomalies

Clinical applications of 3D echocardiography

  • Structure and morphology of heart defects

  • Deformation imaging

  • Volumetry: left ventricle, right ventricle, muscle mass

  • Colour 3D and 3D colour angiography

  • Quantification of regurgitation

  • Defect sizing

  • Delineation of functional morphology of valves including Ebstein's anomaly and atrioventricular septal defect

  • Septation of complex defects

  • Cardiac catheter interventions (defect sizing, catheter manipulation)

Important milestones in real-time 3D echocardiography

  • 1950: M mode imaging

  • 1961: Baum and Greenwood introduce early concept of 3D imaging of the orbit

  • 1970: 2D echocardiography

  • 1974: Dekker and colleagues construct 3D model of the heart using mechanical spatial locator

  • 1977: Matsumoto describes stereoscopic display of a wire frame model of cardiac chambers

  • 1986: Martin and associates use micromanipulator controlled 3D transoesophageal echocardiography (TOE)

  • 1989: Raqueno reconstructs conventional 2D colour flow Doppler images

  • 1990: Von Ramm and Smith use real-time volumetric 3D matrix array probe

  • 1991: Kuroda describes 3D system that rotated the TOE probe

  • 2002: Real-time 3D echocardiography available for clinical use

Practical points for 3D image acquisition

  • Probe position: perpendicular to the structure under evaluation

  • Gain setting: high gain aiming for uniform echogenicity of the structure under evaluation

  • Controls adjusted to get the best blood tissue separation

  • Image centreing at appropriate elevation and sector width

  • Full visualisation of the structure of importance in two orthogonal planes

  • Avoid movement artefacts and use synchronisation with ECG and respiration

  • Acquire multiple full volume 3D loops in colour and grey scale from same window

  • 3D zoom for transoesophageal echocardiography with biplane width adjustment

You can get CPD/CME credits for Education in Heart

Education in Heart articles are accredited by both the UK Royal College of Physicians (London) and the European Board for Accreditation in Cardiology—you need to answer the accompanying multiple choice questions (MCQs). To access the questions, click on BMJ Learning: Take this module on BMJ Learning from the content box at the top right and bottom left of the online article. For more information please go to:

▶ RCP credits: Log your activity in your CPD diary online (—pass mark is 80%.

▶ EBAC credits: Print out and retain the BMJ Learning certificate once you have completed the MCQs—pass mark is 60%. EBAC/ EACCME Credits can now be converted to AMA PRA Category 1 CME Credits and are recognised by all National Accreditation Authorities in Europe (

Please note: The MCQs are hosted on BMJ Learning—the best available learning website for medical professionals from the BMJ Group. If prompted, subscribers must sign into Heart with their journal's username and password. All users must also complete a one-time registration on BMJ Learning and subsequently log in (with a BMJ Learning username and password) on every visit.


  1. A good review on the clinical application of 3D echocardiography in congenital heart disease.
  2. The concept of the colour echocardiographic angiogram is explained here.
  3. This article helps to differentiate between tricuspid valve dysplasia and Ebstein's anomaly.
  4. A detailed review of atrioventricular septal defects and discussion of 3D anatomy.
  5. The complexity of alignment of the ventricular septum and common AV valve components may be understood better by reading this paper.
  6. This article illustrates the difference in the mitral valve annular function in children compared to adults.
  7. A comparative study on 3DE vs MRI for LV volumetry.
  8. A good paper for understanding the role of 3D in CRT.
  9. A source of reference values for normal 3D indices.
View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Competing interests In compliance with EBAC/EACCME guidelines, all authors participating in Education in Heart have disclosed potential conflicts of interest that might cause a bias in the article. JJV is author of the website

  • Provenance and peer review Commissioned; internally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.