- Email: [email protected]
Evaluation of left ventricular systolic function by 3D echocardiography: a comparative study with X-ray angiography and radionuclide angiography
European Journal of Ultrasound 11 (2000) 105 – 115 www.elsevier.com/locate/ejultrasou
Clinical Science: Original paper
Evaluation of left ventricular systolic function by 3D echocardiography: a comparative study with X-ray angiography and radionuclide angiography Ve´ronique Eder a,*, Ste´phane He´rault b, Charles Hudelo c, Bruno Giraudeau d, Christian Marchal a, Laurent Quilliet c, Jean-Marie Pottier a, Philippe Arbeille a a b
Ser6ice Me´decine Nucle´aire et Ultrasons, CHU Trousseau, 37044 Tours, France Unite´ de Me´decine et Physiologie Spatiale, CHU Trousseau, 37044 Tours, France c Ser6ice de Cardiologie A, CHU Trousseau, 37044 Tours, France d Centre de Recherche Clinique, Faculte´ de Me´decine, 37032 Tours, France
Received 12 August 1999; received in revised form 20 January 2000; accepted 20 January 2000
Abstract Objecti6e: the aim of this study was to evaluate left ventricular systolic function by 3D ultrasound as compared to with radionuclide and X-ray angiographies. Methods: one hundred and four patients were examinated by 3D ultrasound (3D-US) but only 72 examinations were successful. Thirty patients were investigated by 3D-US, M-mode US or bidimensional (2D) US, and X-ray angiography (group I) and 42 patients were investigated by 3D-US, M-mode, or 2D, and radionuclide angiography (group II). Results: the correlation between ejection fraction (EF) evaluated by 3D-US and reference methods was found to be good and similar for the two groups (r =0.75; P B 10 − 4 for group I and r= 0.76; P B10 − 4 for group II). The correlation between EF calculated by conventional 2D-US and by reference methods was lower (r=0.60; P= 0.04 for group I and r= 0.54; P = 0.001 for group II). The correlation between EF evaluated by 3D- and 2D-US was modest (r= 0.55; P =0.001 for the whole group). The correlation between 3D-US left ventricle end-diastolic volume (EDV) and end-systolic volume (ESV) and those evaluated by X-ray angiography was also modest (r=0.33; NS for EDV and r= 0.60; PB 10 − 4 for ESV). The correlations between EDV and ESV in 3D-US, and those evaluated from radionuclide angiography were fairly good and in the same range (r= 0.76; PB10 − 4 and r=0.87; P B 10 − 4). Conclusion: the 3D-US system using a rotating probe in an apical view is valuable for evaluation of left ventricular systolic function. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: 3D-Ultrasound; Left ventricular ejection fraction; Radionuclide angiography; X-ray ventriculography
* Corresponding author. Tel.: + 33-2-47475939; fax: +33-2-47475913. E-mail address: [email protected] (V. Eder) 0929-8266/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 9 2 9 - 8 2 6 6 ( 0 0 ) 0 0 0 7 7 - X
106
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
1. Introduction Bidimensional (2D)- and time motion (Mmode) echocardiography can be used to assess left ventricular systolic function by ejection fraction (EF) determination. However, the calculation of ventricular volumes is based on geometric assumptions and the results lack accuracy, in particular in the case of left ventricle regional wall motion abnormalities. Thus, X-ray angiography and radionuclide angiography remain the gold standard methods for left EF determination. Radionuclide angiography is a much more precise method and has a predictive value for prognosis determination
(Bonow, 1994). In the case of tomographic acquisition it allows a very precise estimation of regional wall kinetic function (Lee et al., 1998). The limitations of this method are the cost value and the delivered irradiation. X-ray angiography is a routine reference method for left ventricular fraction calculation. The biplane acquisition allows more precise calculation than monoplane acquisition. With digital substraction, the use of lower doses of contrast agent is possible (Shiba et al., 1991). Nevertheless it remains an invasive study and tomographic acquisition cannot be performed. 3D-US acquisition systems have been recently developed for the calculation of the ventricle vol-
Fig. 1. Volume reconstruction of the left ventricle volume at the end-diastole (bottom right). Two outlines are drawn manually (top left and top right) and the third is automatically displayed on the last image (bottom left).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
107
Fig. 2. Comparison of ejection fraction measured by 3D-echography and X-ray ventriculography (group I).
umes and ejection fraction (Nosir et al., 1996; Pini et al., 1997; Acar et al., 1998; Mele et al., 1998; Rodevand et al., 1998). The aim of this study was to evaluate feasability and accuracy of a commercial 3D echocardiographic system for left ventricle volume and ejection fraction in comparison with X-ray angiography and radionuclide angiography.
2. Methods
2.1. 3D echocardiographic method The 3D-US device consisted of the System V (Vingmed Medical, Trondheim, Norway) with two probes using conventional 2D ultrasound transducers tilting or rotating inside the probe head.
The tilting probe was used in parasternal incidences and collected a set of successive views showing transverse views of the ventricle between the mitral valve and the apex. The rotating probe was used in a four-chamber incidence and collected a set of coaxial views showing apical views of the ventricle, including mitral valve and apex, and oriented around the long ventricle axis. The acquisition was ECG gated. Each rotating or tilting plane was acquired during one cardiac cycle. The recorded view number depended on the rotating or tilting angle between each plane. However, the memory capacity could not accept more than six planes. The acquisition procedure had to be repeated several times in order to find the apical or parasternal incidence which provided views of good quality and containing the limits of the ventricle. These views were stored
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
108
into the memory of the device and then processed using the Echopac workstation. Three apical views of the ventricle were displayed on the screen in real time throughout the period of recording as a cineloop presentation. The digitized ECG was used for selecting the end-diastolic and the end-systolic views of this dynamic sequence. After contouring of the ventricle outlines on each of the three end-diastolic views, the system extrapolated the left ventricle end-diastolic volume (EDV), or end-systolic volume (ESV) and displayed it in a 4th window of the screen (Fig. 1). The calculated volumes (ESV) and (EDV) were expressed in ml. The stroke volume (SV = EDV −ESV), and the ejection fraction (EF= (EDV −ESV)/EDV × 100) were calculated from these values.
2.2. References methods 2.2.1. Echography M-mode and 2D When no regional wall motion abnormality
was present, the ventricle volume was evaluated by using the simplified Teicholtz formula (V= 7D 3/2,4D) where D is the end-diastolic diameter or end-systolic diameter measured in long parasternal axis in M-mode. In other case the ejection fraction was estimated by the index as described by Berning et al. (1992) using bidimensionnal mode.
2.3. Radionuclide angiography Cardiac isotopic examination was based on the intravenous injection of 20 mCi (740MBq) of serum albumin labelled with technetium 99m. The acquisition was ECG gated. Planar views were performed at several incidences by a twohead gamma-camera (DST, Sopha Medical Vision). The ejection fraction was calculated on the 45° left oblique view. Left volume calculation was approximated by the radioactivity counting rate on a region of interest corresponding to the
Fig. 3. Comparison of end-diastolic volume measured by 3D-echography and X-ray ventriculography (group I).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
109
Fig. 4. Comparison of end-systolic volume measured by 3D-echography and X-ray ventriculography (group I).
left ventricle at the end-diastole (EDC) and at the end-systole (ESC). Left ventricular ejection fraction was automatically calculated from these values (Hains et al., 1987).
2.4. X-ray angiography (6entriculography) This method was based on the intraventricular injection of X-ray contrast agent (25 ml of Xenetix®, 350 mg/ml, 9 ml/s) and exposure to X-rays under two orthogonal longitudinal ventricle incidences. The ventricle volume was calculated from the ventricle outlines on these two planes assuming that the cross section of the ventricle was circular or elliptical at any level along the long axis. The two incidences were collected in one shot on two different cardiac cycles. Thus there were no multiple acquisitions
on each incidence. EF was calculated from these volumes (Gault, 1975).
2.5. Material The population consisted of patients hospitalized for cardiological problems at the University hospital and whose underwent isotopic or X-ray angiography for left ventricular systolic function evaluation. The patients were asked to accept or not an additional ultrasound 3D examination. One hundred and four were examinated by 3D-US. The acquisition was successful for only 70% of the examinations; ECG gating failed, for the other cases. When the acquisition was effective, the quality of images was good for left ventricular outline delimitation for all the pa-
110
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
tients with the rotating system, but was never quite good enough with the tilting probe. Only rotating probe data were processed. Thus, 31 patients (23 male and eight female, age mean of 639 8 years) were investigated by 3D-US, conventional echography, and X-ray angiography (group I) and 42 patients (33 male and nine female, age mean of 64 99 years) were investigated by 3D-US, conventional echography, and radionuclide angiogra phy (group II). Seven patients (22%) in group I and eight patients (19%) in group II had ischaemic cardiomyopathy with myocardial infarction history and regional wall motion abnormalities. The other patients underwent the exploration for valular (n = 12), dilated cardiomyopathy (n =35), ischaemic cardiomyopathy (n= 16) or hypertrophic cardiomyopathy (n= 5) and had an homogeneous left ventricular contraction. All patients were in sinus rhythm.
2.6. Statistical methods The correlation coefficient between the values of EDV, EDV, EF calculated by the different methods was tested with Spearmann rank correlation analysis. The confidence interval (CI) 95% was estimated by Fisher transformation. Bland and Altman (1986) analysis was made for comparison of the ejection fraction obtained with 3D-US and the reference methods.
3. Results The results of the measurements are shown in Figs. 2–7. The mean ejection fraction value was 62% (range 23–96; SD 19%) for group I and 47% (range 14–90; SD 18%) for group II.
Fig. 5. Comparison of ejection fraction measured by 3D-echography and radionuclide angiography (group II).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
111
Fig. 6. Comparison of end-diastolic volume measured by 3D-echography and end-diastolic counting rate measured by radionuclide angiography (group II).
The correlation between EF evaluated from 3D-US and X-ray angiography or radionuclide angiography was found good and similar for the two groups (r=0.75, CI 95%=[0.53; 0.87], PB 10-4 group I; r=0.76, CI 95%=[0.59; 0.86], PB 10-4 group II). The correlation between EF determined with TM or 2D echography and the reference methods was less than with 3D-US (r= 0.60, CI 95%=[0.04; 0.87], P= 0.04 groupI; r= 0.54, CI 95%=[0.24; 0.74], P=0.001 group II). The correlation between EF evaluated by 3Dand 2D-US was modest (r=0.55; P= 0.001 for the whole group). Bland and Altman analysis shows that the correlation remained the same for low and for high ejection fraction value in both groups (Figs. 8 and 9).
The correlation between EDV and ESV evaluated from 3D-US and those evaluated by X-ray ventriculography was in the same range (respectively, r=0.33, CI 95%= [ − 0.03; 0.61], NS and r= 0.60, CI 95%= [0.32; 0.79], pB 10 − 4). The correlation between EDV and ESV evaluated from 3D-US and radionuclide angiography was better (respectively r=0.76, CI 95%= [0.59; 0.86], PB 10 − 4 and r= 0.87, CI 95%= [0.77; 0.93], PB10 − 4).
4. Discussion The correlation between EF calculated from 3D-US and EF calculated from both isotopic and X-ray angiographies was found to be of high
112
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
Fig. 7. Comparison of end-systolic volume measured by 3D-echography and end-systolic counting rate measured by radionuclide angiography (group II).
value for routine cardiological applications. The correlation between EF calculated from 3D-US and EF calculated by conventional echo-TM or 2D was more modest. 3D echography is thus a much more precise method. There was a fairly high correlation between volumes measured by 3D-US and volumes calculated from g-ray activity on isotopic angiographies despite a lack of calibration. The correlation between the volumes measured by 3D-US and those calculated from X-ray angiography is not high. Nevertheless even the absolute volume values are not well correlated with the X-ray ventriculography, the ratio of these volumes (like EF) can be accurately determined. However, Bland and Altman analysis showed that the systematic errors were the same for the two groups. This result is surprising because the normal ejection values are not the same for Xray and isotopıˆc angiography. This would be ex-
plained by the operator intervention in the determination of ventricular cavity contours which could be influenced by visual ejection fraction appreciation.
Fig. 8. Bland and Altman graphs for ejection fraction measured by isotopic angiography and 3D echocardiography.
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
Fig. 9. As in Fig. 8, now for X-ray ventriculography and 3D echocardiography.
With the rotating probe, the 3D-US procedure is based on the acquisition of several incidences starting from an original four chambers apical view, all of them containing the ventricle long axis. Thus the probe stays fairly perpendicular to the skin and interspace ribs and the ultrasound windows remain practically the same in all incidences. Conversely, in case of the tilt probe, the successive echographic views are obtained through different orientations of the probe head, thus the ultrasound windows are not similar in all the incidences. This would explain why it was very difficult to record cardiac images of high quality for each incidence. In addition the rotation or tilting of the probes were gated with the ECG in order to ensure that each incidence is investigated during at least one cardiac cycle, which guaranteed that the whole left ventricle was investigated both in systole and diastole. However, this process can induce errors in the volume calculation. For instance any arrhythmia, cardiac low heart rate, repiratory or physical movement of the patient could change the position of the heart and thus induced errors on the ultrasound image. Moreover the number of recorded plans is limited by the time duration of the examination and by the memory capacity of the computer. The cardiac volume, presented in 3D, is reconstructed from the contours delineated on the 2D planes stored into the memory during the acquisition phase. Due to the limited number of planes,
113
a rather large proportion of the myocardium is not seen by the ultrasound beam. Thus, several parts of the heart cannot be displayed because theire images are not stored into the memory. As a consequence this method does not allow to study the segmental contraction of the myocardium. Volume reconstruction was initiated with two manual drawings of left ventricular contour and was successfull for all patients. The following ventricular outlines was automatically calculated and traced by the Echopac algorithm. Unfortunatly these contours were always badly fitted with the real cavity border, and had to be manually modified. The reconstruction algorithm would be improved. Our results are in agreement with those of previous studies but not so good. Altmann et al. (1997) with a 2D-US tilted probe found a very high correlation (r=0.98) between the 3D and the magnetic resonance imaging data for ventricle volume and EF. Gopal et al. (1995) calculated left ventricular volume from short axis views, and found a high correlation for ejection fraction when comparing with radionuclide angiography (r= 0.94) or with cine-angiography (r= 0.82). Nosir et al. (1996) found also a very high degree of correlation. Their method used many transverse cross sections of the left ventricle. Acar et al. (1998) used a rotating system but the left ventricular volumes were calculated using Simpson’s rule with 12 slices. They found a high correlation with radionuclide angiography in population of children. Mele et al. (1998) used also a rotating system and found better results with 3D than with 2D echography. De Castro et al. (1998) found 3D echography particulary interesting for cardiac surgery. Shiota et al. (1998) and Von Ramm and Smith (1990) used a matrix probe which allowed 3D realtime acquisition. Thus all the echographic data from the whole heart could be stored in the memory of the computer in one cardiac cycle. They found a good correlation (r= 0.80) between right ventricle stroke volume changes measured by 3D-US acquisiton and with an electromagnetic flow meter on an animal model. Pini et al. (1997) used 3D echography for right ventricular function evaluation.
114
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
The major clinical advantage of the 3D echocardiography is that it allows repeated assessment of the ejection fraction because it is a non invasive method contrary to isotopic or X-ray angiographies. Nevertheless the isotopic g angiography provides more informations than EF such as diastolic function and regional wall motion evaluation when tomography is performed contrary to the present 3D system. The performances of the 3D method will probably improve in the near future. If the scan head of the tilting probe could be reduced significantly, the tilting probes would become operational and would perform a more accurately determination of the ventricle (Rodevand et al., 1998). More operational automatic contouring of the left ventricle limits would make the processing step much quicker. New ‘active contour’ algorithms are still under development but for 2D-US images processing only. 3D echocardiography would be useful for the clinician as a non invasive assessment of the ejection fraction, but also for the investigation of the mitral or aortic valves (Salustri et al., 1996; Dall’Agata et al., 1999) and for the intraventricular flows by 3D color Doppler mapping (Breburda et al., 1998; De Simone et al., 1999).
4.1. Study limitations Patients were included successively and a part of patients did not have a left systolic dysfunction. Estimation of EF was made with M-mode measurements because our experimental device did not allow us to use a standard 2D-US method.
References Acar P, Maunoury C, Antonietti T, Bonnet D, Sidi D, Kachaner J. Left ventricular ejection fraction in children measured by three-dimensional echocardiography using a new tranthoracic integrated 3D-probe. A comparison with equilibrium radionuclide angiography. Eur Heart J 1998;19:1583–8. Altmann K, Shen Z, Boxt LM, et al. Comparison of three-dimensional echocardiographic assessment of volume, mass, and function in children with functionally single left ven.
tricles with two-dimensional echocardiography and magnetic resonance imaging. Am J Cardiol 1997;80:1060–5. Berning J, Rokkedal Nielsen J, Launbjerg J, Fogh J, Mickley H, Andersen PE. Rapid estimation of left ventricular ejection fraction in acute myocardial infarction by echocardiographic wall motion analysis. Cardiology 1992;80:257– 66. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307 – 10. Bonow RO. Prognostic assessment in coronary artery disease: role of radionuclide angiography. J Nucl Cardiol 1994;1:280 – 91. Breburda CS, Griffin BP, Pu M, Rodriguez L, Cosgrove DM III, Thomas JD. Three-dimensional echocardiography planimetry of maximal regurgitant orifice area in myxomatous mitral regurgitation: intraoperative comparison with proximal flow convergence. J Am Coll Cardiol 1998;32:432 – 7. Dall’Agata A, Cromme-Dijkhuis AH, Meijboom FJ, et al. Use of three-dimensional echocardiography for analysis of outflow obstruction in congenital heart disease. Am J Cardiol 1999;83:921 – 3. De Castro S, Yao J, Pandian NG. Three-dimensional echocardiography: clinical relevance and application. Am J Cardiol 1998;81:96 – 102. De Simone R, Glombitza G, Vahl CF, Albers J, Meinzer HP, Hagl S. Three-dimensional color Doppler: a new approach for quantitative assessment of mitral regurgitant jets. J Am Soc Echocardiogr 1999;12:173 – 85. Gault JH. Angiographic estimation of left ventricular volume. Cathet Cardiovasc Diagn 1975;1:7 – 16. Gopal AS, Shen Z, Sapin PM, et al. Assessment of cardiac function by three-dimensional echocardiography compared with conventional noninvasive methods, Circulation 1995;92:842 – 853. Hains AD, Khawaja IA, Lahiri A, Raftery FB. Radionuclide left ventricular ejection fraction: a comparison of three methods. Br Heart J 1987;57:232 – 6. Lee HS, Cross S, Norton M, Walton S. Comparison between planar and tomographic radionuclide ventriculography for detecting inferior wall motion abnormalities. Clin Radiol 1998;53:264. Mele D, Maehle J, Pedini I, Alboni P, Levine RA. Three-dimensional echocardiographic reconstruction: description and applications of a simplified technique for quantitative assessment of left ventricular size and function, Am J Cardiol 1998;81:107 – 110. Nosir YF, Fioretti PM, Vletter WB, et al. Accurate measurement of left ventricular ejection fraction by three-dimensional echocardiography. A comparison with radionuclide angiography. Circulation 1996;94:460 – 6. Pini R, Giannazzo G, Di Bari M, et al. Transthoracic threedimensional echocardiographic reconstruction of left and right ventricles: in vitro validation and comparison with magnetic resonance imaging. Am Heart J 1997;133:221– 9.
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115 Rodevand O, Bjornerheim R, Aakhus S, Kjekshus J. Left ventricular volumes assessed by different new three-dimensional echocardiographic methods and ordinary biplane technique. Int J Card Imaging 1998;14:55 – 63. Salustri A, Becker AE, van Herwerden L, Vletter WB, Ten Cate FJ, Roelandt JR. Three-dimensional echocardiography of normal and pathologic mitral valve: a comparison with two-dimensional transesophageal echocardiography. J Am Coll Cardiol 1996;27:1502–10. Shiba K, Nanki M, Hori H, Sakamoto N. Evaluation of the
.
115
densitometric method of ejection fraction by direct digital subtraction left ventriculography. Cathet Cardiovasc Diagn 1991;23:61 – 6. Shiota T, Jones M, Chikada M, et al. Real-time three-dimensional echocardiography for determining right ventricular stroke volume in an animal model of chronic right ventricular volume overload, Circulation 1998;97:1897– 1900. Von Ramm OT, Smith SW. Real time volumetric ultrasound imaging system. J Digit Imaging 1990;3:261 – 6.
Clinical Science: Original paper
Evaluation of left ventricular systolic function by 3D echocardiography: a comparative study with X-ray angiography and radionuclide angiography Ve´ronique Eder a,*, Ste´phane He´rault b, Charles Hudelo c, Bruno Giraudeau d, Christian Marchal a, Laurent Quilliet c, Jean-Marie Pottier a, Philippe Arbeille a a b
Ser6ice Me´decine Nucle´aire et Ultrasons, CHU Trousseau, 37044 Tours, France Unite´ de Me´decine et Physiologie Spatiale, CHU Trousseau, 37044 Tours, France c Ser6ice de Cardiologie A, CHU Trousseau, 37044 Tours, France d Centre de Recherche Clinique, Faculte´ de Me´decine, 37032 Tours, France
Received 12 August 1999; received in revised form 20 January 2000; accepted 20 January 2000
Abstract Objecti6e: the aim of this study was to evaluate left ventricular systolic function by 3D ultrasound as compared to with radionuclide and X-ray angiographies. Methods: one hundred and four patients were examinated by 3D ultrasound (3D-US) but only 72 examinations were successful. Thirty patients were investigated by 3D-US, M-mode US or bidimensional (2D) US, and X-ray angiography (group I) and 42 patients were investigated by 3D-US, M-mode, or 2D, and radionuclide angiography (group II). Results: the correlation between ejection fraction (EF) evaluated by 3D-US and reference methods was found to be good and similar for the two groups (r =0.75; P B 10 − 4 for group I and r= 0.76; P B10 − 4 for group II). The correlation between EF calculated by conventional 2D-US and by reference methods was lower (r=0.60; P= 0.04 for group I and r= 0.54; P = 0.001 for group II). The correlation between EF evaluated by 3D- and 2D-US was modest (r= 0.55; P =0.001 for the whole group). The correlation between 3D-US left ventricle end-diastolic volume (EDV) and end-systolic volume (ESV) and those evaluated by X-ray angiography was also modest (r=0.33; NS for EDV and r= 0.60; PB 10 − 4 for ESV). The correlations between EDV and ESV in 3D-US, and those evaluated from radionuclide angiography were fairly good and in the same range (r= 0.76; PB10 − 4 and r=0.87; P B 10 − 4). Conclusion: the 3D-US system using a rotating probe in an apical view is valuable for evaluation of left ventricular systolic function. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: 3D-Ultrasound; Left ventricular ejection fraction; Radionuclide angiography; X-ray ventriculography
* Corresponding author. Tel.: + 33-2-47475939; fax: +33-2-47475913. E-mail address: [email protected] (V. Eder) 0929-8266/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 9 2 9 - 8 2 6 6 ( 0 0 ) 0 0 0 7 7 - X
106
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
1. Introduction Bidimensional (2D)- and time motion (Mmode) echocardiography can be used to assess left ventricular systolic function by ejection fraction (EF) determination. However, the calculation of ventricular volumes is based on geometric assumptions and the results lack accuracy, in particular in the case of left ventricle regional wall motion abnormalities. Thus, X-ray angiography and radionuclide angiography remain the gold standard methods for left EF determination. Radionuclide angiography is a much more precise method and has a predictive value for prognosis determination
(Bonow, 1994). In the case of tomographic acquisition it allows a very precise estimation of regional wall kinetic function (Lee et al., 1998). The limitations of this method are the cost value and the delivered irradiation. X-ray angiography is a routine reference method for left ventricular fraction calculation. The biplane acquisition allows more precise calculation than monoplane acquisition. With digital substraction, the use of lower doses of contrast agent is possible (Shiba et al., 1991). Nevertheless it remains an invasive study and tomographic acquisition cannot be performed. 3D-US acquisition systems have been recently developed for the calculation of the ventricle vol-
Fig. 1. Volume reconstruction of the left ventricle volume at the end-diastole (bottom right). Two outlines are drawn manually (top left and top right) and the third is automatically displayed on the last image (bottom left).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
107
Fig. 2. Comparison of ejection fraction measured by 3D-echography and X-ray ventriculography (group I).
umes and ejection fraction (Nosir et al., 1996; Pini et al., 1997; Acar et al., 1998; Mele et al., 1998; Rodevand et al., 1998). The aim of this study was to evaluate feasability and accuracy of a commercial 3D echocardiographic system for left ventricle volume and ejection fraction in comparison with X-ray angiography and radionuclide angiography.
2. Methods
2.1. 3D echocardiographic method The 3D-US device consisted of the System V (Vingmed Medical, Trondheim, Norway) with two probes using conventional 2D ultrasound transducers tilting or rotating inside the probe head.
The tilting probe was used in parasternal incidences and collected a set of successive views showing transverse views of the ventricle between the mitral valve and the apex. The rotating probe was used in a four-chamber incidence and collected a set of coaxial views showing apical views of the ventricle, including mitral valve and apex, and oriented around the long ventricle axis. The acquisition was ECG gated. Each rotating or tilting plane was acquired during one cardiac cycle. The recorded view number depended on the rotating or tilting angle between each plane. However, the memory capacity could not accept more than six planes. The acquisition procedure had to be repeated several times in order to find the apical or parasternal incidence which provided views of good quality and containing the limits of the ventricle. These views were stored
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
108
into the memory of the device and then processed using the Echopac workstation. Three apical views of the ventricle were displayed on the screen in real time throughout the period of recording as a cineloop presentation. The digitized ECG was used for selecting the end-diastolic and the end-systolic views of this dynamic sequence. After contouring of the ventricle outlines on each of the three end-diastolic views, the system extrapolated the left ventricle end-diastolic volume (EDV), or end-systolic volume (ESV) and displayed it in a 4th window of the screen (Fig. 1). The calculated volumes (ESV) and (EDV) were expressed in ml. The stroke volume (SV = EDV −ESV), and the ejection fraction (EF= (EDV −ESV)/EDV × 100) were calculated from these values.
2.2. References methods 2.2.1. Echography M-mode and 2D When no regional wall motion abnormality
was present, the ventricle volume was evaluated by using the simplified Teicholtz formula (V= 7D 3/2,4D) where D is the end-diastolic diameter or end-systolic diameter measured in long parasternal axis in M-mode. In other case the ejection fraction was estimated by the index as described by Berning et al. (1992) using bidimensionnal mode.
2.3. Radionuclide angiography Cardiac isotopic examination was based on the intravenous injection of 20 mCi (740MBq) of serum albumin labelled with technetium 99m. The acquisition was ECG gated. Planar views were performed at several incidences by a twohead gamma-camera (DST, Sopha Medical Vision). The ejection fraction was calculated on the 45° left oblique view. Left volume calculation was approximated by the radioactivity counting rate on a region of interest corresponding to the
Fig. 3. Comparison of end-diastolic volume measured by 3D-echography and X-ray ventriculography (group I).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
109
Fig. 4. Comparison of end-systolic volume measured by 3D-echography and X-ray ventriculography (group I).
left ventricle at the end-diastole (EDC) and at the end-systole (ESC). Left ventricular ejection fraction was automatically calculated from these values (Hains et al., 1987).
2.4. X-ray angiography (6entriculography) This method was based on the intraventricular injection of X-ray contrast agent (25 ml of Xenetix®, 350 mg/ml, 9 ml/s) and exposure to X-rays under two orthogonal longitudinal ventricle incidences. The ventricle volume was calculated from the ventricle outlines on these two planes assuming that the cross section of the ventricle was circular or elliptical at any level along the long axis. The two incidences were collected in one shot on two different cardiac cycles. Thus there were no multiple acquisitions
on each incidence. EF was calculated from these volumes (Gault, 1975).
2.5. Material The population consisted of patients hospitalized for cardiological problems at the University hospital and whose underwent isotopic or X-ray angiography for left ventricular systolic function evaluation. The patients were asked to accept or not an additional ultrasound 3D examination. One hundred and four were examinated by 3D-US. The acquisition was successful for only 70% of the examinations; ECG gating failed, for the other cases. When the acquisition was effective, the quality of images was good for left ventricular outline delimitation for all the pa-
110
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
tients with the rotating system, but was never quite good enough with the tilting probe. Only rotating probe data were processed. Thus, 31 patients (23 male and eight female, age mean of 639 8 years) were investigated by 3D-US, conventional echography, and X-ray angiography (group I) and 42 patients (33 male and nine female, age mean of 64 99 years) were investigated by 3D-US, conventional echography, and radionuclide angiogra phy (group II). Seven patients (22%) in group I and eight patients (19%) in group II had ischaemic cardiomyopathy with myocardial infarction history and regional wall motion abnormalities. The other patients underwent the exploration for valular (n = 12), dilated cardiomyopathy (n =35), ischaemic cardiomyopathy (n= 16) or hypertrophic cardiomyopathy (n= 5) and had an homogeneous left ventricular contraction. All patients were in sinus rhythm.
2.6. Statistical methods The correlation coefficient between the values of EDV, EDV, EF calculated by the different methods was tested with Spearmann rank correlation analysis. The confidence interval (CI) 95% was estimated by Fisher transformation. Bland and Altman (1986) analysis was made for comparison of the ejection fraction obtained with 3D-US and the reference methods.
3. Results The results of the measurements are shown in Figs. 2–7. The mean ejection fraction value was 62% (range 23–96; SD 19%) for group I and 47% (range 14–90; SD 18%) for group II.
Fig. 5. Comparison of ejection fraction measured by 3D-echography and radionuclide angiography (group II).
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
111
Fig. 6. Comparison of end-diastolic volume measured by 3D-echography and end-diastolic counting rate measured by radionuclide angiography (group II).
The correlation between EF evaluated from 3D-US and X-ray angiography or radionuclide angiography was found good and similar for the two groups (r=0.75, CI 95%=[0.53; 0.87], PB 10-4 group I; r=0.76, CI 95%=[0.59; 0.86], PB 10-4 group II). The correlation between EF determined with TM or 2D echography and the reference methods was less than with 3D-US (r= 0.60, CI 95%=[0.04; 0.87], P= 0.04 groupI; r= 0.54, CI 95%=[0.24; 0.74], P=0.001 group II). The correlation between EF evaluated by 3Dand 2D-US was modest (r=0.55; P= 0.001 for the whole group). Bland and Altman analysis shows that the correlation remained the same for low and for high ejection fraction value in both groups (Figs. 8 and 9).
The correlation between EDV and ESV evaluated from 3D-US and those evaluated by X-ray ventriculography was in the same range (respectively, r=0.33, CI 95%= [ − 0.03; 0.61], NS and r= 0.60, CI 95%= [0.32; 0.79], pB 10 − 4). The correlation between EDV and ESV evaluated from 3D-US and radionuclide angiography was better (respectively r=0.76, CI 95%= [0.59; 0.86], PB 10 − 4 and r= 0.87, CI 95%= [0.77; 0.93], PB10 − 4).
4. Discussion The correlation between EF calculated from 3D-US and EF calculated from both isotopic and X-ray angiographies was found to be of high
112
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
Fig. 7. Comparison of end-systolic volume measured by 3D-echography and end-systolic counting rate measured by radionuclide angiography (group II).
value for routine cardiological applications. The correlation between EF calculated from 3D-US and EF calculated by conventional echo-TM or 2D was more modest. 3D echography is thus a much more precise method. There was a fairly high correlation between volumes measured by 3D-US and volumes calculated from g-ray activity on isotopic angiographies despite a lack of calibration. The correlation between the volumes measured by 3D-US and those calculated from X-ray angiography is not high. Nevertheless even the absolute volume values are not well correlated with the X-ray ventriculography, the ratio of these volumes (like EF) can be accurately determined. However, Bland and Altman analysis showed that the systematic errors were the same for the two groups. This result is surprising because the normal ejection values are not the same for Xray and isotopıˆc angiography. This would be ex-
plained by the operator intervention in the determination of ventricular cavity contours which could be influenced by visual ejection fraction appreciation.
Fig. 8. Bland and Altman graphs for ejection fraction measured by isotopic angiography and 3D echocardiography.
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
Fig. 9. As in Fig. 8, now for X-ray ventriculography and 3D echocardiography.
With the rotating probe, the 3D-US procedure is based on the acquisition of several incidences starting from an original four chambers apical view, all of them containing the ventricle long axis. Thus the probe stays fairly perpendicular to the skin and interspace ribs and the ultrasound windows remain practically the same in all incidences. Conversely, in case of the tilt probe, the successive echographic views are obtained through different orientations of the probe head, thus the ultrasound windows are not similar in all the incidences. This would explain why it was very difficult to record cardiac images of high quality for each incidence. In addition the rotation or tilting of the probes were gated with the ECG in order to ensure that each incidence is investigated during at least one cardiac cycle, which guaranteed that the whole left ventricle was investigated both in systole and diastole. However, this process can induce errors in the volume calculation. For instance any arrhythmia, cardiac low heart rate, repiratory or physical movement of the patient could change the position of the heart and thus induced errors on the ultrasound image. Moreover the number of recorded plans is limited by the time duration of the examination and by the memory capacity of the computer. The cardiac volume, presented in 3D, is reconstructed from the contours delineated on the 2D planes stored into the memory during the acquisition phase. Due to the limited number of planes,
113
a rather large proportion of the myocardium is not seen by the ultrasound beam. Thus, several parts of the heart cannot be displayed because theire images are not stored into the memory. As a consequence this method does not allow to study the segmental contraction of the myocardium. Volume reconstruction was initiated with two manual drawings of left ventricular contour and was successfull for all patients. The following ventricular outlines was automatically calculated and traced by the Echopac algorithm. Unfortunatly these contours were always badly fitted with the real cavity border, and had to be manually modified. The reconstruction algorithm would be improved. Our results are in agreement with those of previous studies but not so good. Altmann et al. (1997) with a 2D-US tilted probe found a very high correlation (r=0.98) between the 3D and the magnetic resonance imaging data for ventricle volume and EF. Gopal et al. (1995) calculated left ventricular volume from short axis views, and found a high correlation for ejection fraction when comparing with radionuclide angiography (r= 0.94) or with cine-angiography (r= 0.82). Nosir et al. (1996) found also a very high degree of correlation. Their method used many transverse cross sections of the left ventricle. Acar et al. (1998) used a rotating system but the left ventricular volumes were calculated using Simpson’s rule with 12 slices. They found a high correlation with radionuclide angiography in population of children. Mele et al. (1998) used also a rotating system and found better results with 3D than with 2D echography. De Castro et al. (1998) found 3D echography particulary interesting for cardiac surgery. Shiota et al. (1998) and Von Ramm and Smith (1990) used a matrix probe which allowed 3D realtime acquisition. Thus all the echographic data from the whole heart could be stored in the memory of the computer in one cardiac cycle. They found a good correlation (r= 0.80) between right ventricle stroke volume changes measured by 3D-US acquisiton and with an electromagnetic flow meter on an animal model. Pini et al. (1997) used 3D echography for right ventricular function evaluation.
114
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115
The major clinical advantage of the 3D echocardiography is that it allows repeated assessment of the ejection fraction because it is a non invasive method contrary to isotopic or X-ray angiographies. Nevertheless the isotopic g angiography provides more informations than EF such as diastolic function and regional wall motion evaluation when tomography is performed contrary to the present 3D system. The performances of the 3D method will probably improve in the near future. If the scan head of the tilting probe could be reduced significantly, the tilting probes would become operational and would perform a more accurately determination of the ventricle (Rodevand et al., 1998). More operational automatic contouring of the left ventricle limits would make the processing step much quicker. New ‘active contour’ algorithms are still under development but for 2D-US images processing only. 3D echocardiography would be useful for the clinician as a non invasive assessment of the ejection fraction, but also for the investigation of the mitral or aortic valves (Salustri et al., 1996; Dall’Agata et al., 1999) and for the intraventricular flows by 3D color Doppler mapping (Breburda et al., 1998; De Simone et al., 1999).
4.1. Study limitations Patients were included successively and a part of patients did not have a left systolic dysfunction. Estimation of EF was made with M-mode measurements because our experimental device did not allow us to use a standard 2D-US method.
References Acar P, Maunoury C, Antonietti T, Bonnet D, Sidi D, Kachaner J. Left ventricular ejection fraction in children measured by three-dimensional echocardiography using a new tranthoracic integrated 3D-probe. A comparison with equilibrium radionuclide angiography. Eur Heart J 1998;19:1583–8. Altmann K, Shen Z, Boxt LM, et al. Comparison of three-dimensional echocardiographic assessment of volume, mass, and function in children with functionally single left ven.
tricles with two-dimensional echocardiography and magnetic resonance imaging. Am J Cardiol 1997;80:1060–5. Berning J, Rokkedal Nielsen J, Launbjerg J, Fogh J, Mickley H, Andersen PE. Rapid estimation of left ventricular ejection fraction in acute myocardial infarction by echocardiographic wall motion analysis. Cardiology 1992;80:257– 66. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307 – 10. Bonow RO. Prognostic assessment in coronary artery disease: role of radionuclide angiography. J Nucl Cardiol 1994;1:280 – 91. Breburda CS, Griffin BP, Pu M, Rodriguez L, Cosgrove DM III, Thomas JD. Three-dimensional echocardiography planimetry of maximal regurgitant orifice area in myxomatous mitral regurgitation: intraoperative comparison with proximal flow convergence. J Am Coll Cardiol 1998;32:432 – 7. Dall’Agata A, Cromme-Dijkhuis AH, Meijboom FJ, et al. Use of three-dimensional echocardiography for analysis of outflow obstruction in congenital heart disease. Am J Cardiol 1999;83:921 – 3. De Castro S, Yao J, Pandian NG. Three-dimensional echocardiography: clinical relevance and application. Am J Cardiol 1998;81:96 – 102. De Simone R, Glombitza G, Vahl CF, Albers J, Meinzer HP, Hagl S. Three-dimensional color Doppler: a new approach for quantitative assessment of mitral regurgitant jets. J Am Soc Echocardiogr 1999;12:173 – 85. Gault JH. Angiographic estimation of left ventricular volume. Cathet Cardiovasc Diagn 1975;1:7 – 16. Gopal AS, Shen Z, Sapin PM, et al. Assessment of cardiac function by three-dimensional echocardiography compared with conventional noninvasive methods, Circulation 1995;92:842 – 853. Hains AD, Khawaja IA, Lahiri A, Raftery FB. Radionuclide left ventricular ejection fraction: a comparison of three methods. Br Heart J 1987;57:232 – 6. Lee HS, Cross S, Norton M, Walton S. Comparison between planar and tomographic radionuclide ventriculography for detecting inferior wall motion abnormalities. Clin Radiol 1998;53:264. Mele D, Maehle J, Pedini I, Alboni P, Levine RA. Three-dimensional echocardiographic reconstruction: description and applications of a simplified technique for quantitative assessment of left ventricular size and function, Am J Cardiol 1998;81:107 – 110. Nosir YF, Fioretti PM, Vletter WB, et al. Accurate measurement of left ventricular ejection fraction by three-dimensional echocardiography. A comparison with radionuclide angiography. Circulation 1996;94:460 – 6. Pini R, Giannazzo G, Di Bari M, et al. Transthoracic threedimensional echocardiographic reconstruction of left and right ventricles: in vitro validation and comparison with magnetic resonance imaging. Am Heart J 1997;133:221– 9.
V. Eder et al. / European Journal of Ultrasound 11 (2000) 105–115 Rodevand O, Bjornerheim R, Aakhus S, Kjekshus J. Left ventricular volumes assessed by different new three-dimensional echocardiographic methods and ordinary biplane technique. Int J Card Imaging 1998;14:55 – 63. Salustri A, Becker AE, van Herwerden L, Vletter WB, Ten Cate FJ, Roelandt JR. Three-dimensional echocardiography of normal and pathologic mitral valve: a comparison with two-dimensional transesophageal echocardiography. J Am Coll Cardiol 1996;27:1502–10. Shiba K, Nanki M, Hori H, Sakamoto N. Evaluation of the
.
115
densitometric method of ejection fraction by direct digital subtraction left ventriculography. Cathet Cardiovasc Diagn 1991;23:61 – 6. Shiota T, Jones M, Chikada M, et al. Real-time three-dimensional echocardiography for determining right ventricular stroke volume in an animal model of chronic right ventricular volume overload, Circulation 1998;97:1897– 1900. Von Ramm OT, Smith SW. Real time volumetric ultrasound imaging system. J Digit Imaging 1990;3:261 – 6.