D. PALADINI, M. VASSALLO, G. SGLAVO, C. LAPADULA and P. MARTINELLI

Transcription

1 Ultrasound Obstet Gynecol 2006; 27: Published online in Wiley InterScience ( DOI: /uog.2749 The role of spatio-temporal image correlation (STIC) with tomographic ultrasound imaging (TUI) in the sequential analysis of fetal congenital heart disease D. PALADINI, M. VASSALLO, G. SGLAVO, C. LAPADULA and P. MARTINELLI Fetal Cardiology Unit, Department of Gynecology and Obstetrics, University Federico II of Naples, Italy KEYWORDS: congenital heart disease; fetal echocardiography; four-dimensional ultrasound; STIC; tomographic imaging ABSTRACT Objective Spatio-temporal image correlation associated with the tomographic ultrasound imaging mode (TUI- STIC) is a new modality that allows a complete sequential analysis of cardiac structures to be displayed on a single panel by showing all echocardiographic transverse views at the same time. The aims of this study were to identify the best settings for displaying the classic echocardiographic views at different gestational ages and to investigate the role of TUI-STIC in the sequential segmental analysis of complex congenital heart disease (CHD). Methods Four-dimensional volumes from 103 cases of confirmed fetal CHD diagnosed and managed at our referral center were evaluated using TUI-STIC. To select the best interslice distance for adequate display of the central cardiovascular connections, each volume was opened and the TUI mode activated, having as a reference the apical four-chamber view. The number of slices was set at nine. The volume was then scrolled until the most significant echocardiographic views were displayed on the screen windows. Then, if too many windows showed intermediate non-diagnostic views, the slice distance was adjusted finely until all key echocardiographic views showed up in the various windows. The interslice distance was regressed against gestational age and the best-fitting curve was identified. Results A sequential segmental analysis could be shown with TUI-STIC in all cases. A linear regression equation best fitted the correlation between interslice distance and advancing gestational age (r 2 = ), with the mean interslice distance being 2.7 (SD, 0.3) mm at gestational weeks, and 4.0 (SD, 0.4) mm at weeks. These settings allowed a complete sequential analysis in all cases. Conclusions TUI-STIC allows a complete sequential analysis of CHD in the fetus. The most suitable interslice distances for all gestational ages could be identified. These data may be used while adopting this imaging modality in the four-dimensional evaluation of fetal CHD. Copyright 2006 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION In pediatric cardiology, the sequential segmental analysis of the central cardiovascular structures is the recommended approach for the characterization of congenital heart disease (CHD) 1,2, as it allows a clear description of the various coexisting abnormalities in complex heart defects, such as heterotaxy syndromes or complex conotruncal anomalies. These CHDs account for less than 8% of major heart defects in the neonate 3 and for 10 15% of all CHDs detected prenatally 4,5, the difference being determined by a double selection bias (easier ultrasound recognition and higher perinatal mortality rate). A sequential approach involving transverse views from the abdomen to the ductal arches was thus adopted and described several years ago 6, and has been found to be advantageous in the characterization of CHD in the fetus 4,5,7. Four-dimensional echocardiography with spatiotemporal image correlation (STIC) is a relatively new technique that allows the acquisition of cardiac volumes 8. Information extracted from these datasets can then be displayed and analyzed in multiplanar or rendering modes 9. One of the latest developments is that of tomographic ultrasound imaging (TUI). This imaging modality allows one to display on a single panel a variable number of reconstructed two-dimensional sections, as in Correspondence to: Dr D. Paladini, Via Petrarca Naples, Italy ( paladini@unina.it) Accepted: 29 December 2005 Copyright 2006 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

2 556 Paladini et al. computed tomography or magnetic resonance imaging. In particular, it displays parallel slices which are orthogonal to the plane of acquisition. Therefore, if the acquisition is performed orthogonal to the fetal body (for example, starting from the apical or transverse four-chamber view), and if the volume is sufficiently large, all consecutive slices from the transverse abdominal to the high mediastinum with the ductal and aortic arches will be displayed automatically. The rationale of this preliminary study was to assess the role of four-dimensional echocardiography with TUI (TUI-STIC) in the sequential analysis of major CHD detected in the fetus. The end-points of this study were: 1) to assess whether TUI-STIC allows sequential segmental analysis of CHD in the fetus to be shown on a single panel of images, and 2) to identify the correct setting for the display of the classic echocardiographic views at different gestational ages, using as a reference view for volume acquisition the four-chamber view of the fetal heart. METHODS Four-dimensional echocardiography (STIC - GE Medical systems, Kretz Ultrasound, Zipf, Austria) has been used in our referral center since its introduction in August In the 2-year period until August 2005, 201 cases of fetal CHD were diagnosed and analyzed in our center, and in 152 of them one or more STIC volumes were acquired and stored. Forty-nine cases were excluded from the analysis due to: sub-optimal quality of the volume (e.g. fetal movements, maternal obesity, 24 cases), volume not including the ductal arches or the transverse abdominal Figure 1 Four-dimensional echocardiography of a normal fetal heart (29 weeks). The upper left window shows a sagittal view of the fetal thorax; lines correspond to the views shown in the other windows (the dotted line corresponds to the ninth window, which is not displayed). Interslice distance = 4.2 mm. The reference dot identifies the lumen of the descending aorta in all windows. Window 3 (bottom row, right): transverse view of the abdomen, the normal situs, with the aorta (reference dot) and the stomach (St) to the left and the inferior vena cava (IVC) to the right of the midline. Window 2: the vena cava can be seen entering the right atrium. Window 1: posterior four-chamber view. Window : classic four-chamber view, with good visualization of the atria, the atrioventricular plane, the ventricles and the septa. Note the reference dot within the descending aorta. LV, left ventricle; RA, right atrium. Window 1: left outflow tract, with the ascending aorta (Ao) arising from the LV. Window 2: right outflow tract, with the main pulmonary artery exiting from the right ventricle. Window 3: right outflow and three vessels view, with the pulmonary artery (Pa) arising from the right ventricle (RV). The aorta (Ao) is seen to the right of the pulmonary artery and the superior vena cava to the right of the aorta. The arrow indicates the trachea. Window 4: transverse view of the aortic arch (arrowhead), to the left of the trachea.

3 Role of TUI-STIC in CHD 557 view (21 cases), or diagnosis unconfirmed (lost to followup or necropsy not performed, four cases). The remaining 103 cases of confirmed fetal CHD constituted the current study population (Table 1). In each case, the best grayscale or color volume illustrating the heart defect was selected for the study. In each case, before activating the acquisition procedure, care had been taken to acquire the volume using as a reference plane an apical or transverse four-chamber view of the fetal heart, depending on whether demonstration of abnormal anatomy or flows, respectively, was required. Prior to storage of the volumes, the orientation was standardized as follows: after acquisition, the image was rotated 180 on the y- axis (laterally, from side to side) in order to store all volumes as if they had been acquired with the fetus in vertex presentation; as a result, the volumes acquired with a ventral approach (apical four-chamber view) always showed the left ventricle on the left side of the image (Figure 1) and volumes acquired with a right lateral approach (transverse four-chamber view) always had the cardiac apex on the left side of the image, and the left ventricle in the lower part. No other rotation maneuvers were performed prior to storage on the hard disk. Analysis of the volume was performed offline on a personal computer using dedicated software (4D-viewer 5.0, GE Medical systems, Kretz Ultrasound, Zipf, Austria). To select the best interslice distance for showing adequately the central cardiovascular connections, each volume was opened and the TUI display mode was activated, using as a reference the four-chamber view. The number of slices was set at nine. The volume was then scrolled until the most significant echocardiographic views were Table 1 Characteristics of the study population: 103 fetuses with congenital heart disease (CHD) confirmed at necropsy/postnatal echocardiography Defect n % Aortic coarctation/aortic arch interruption Atrioventricular septal defect Cardiomyopathy Common arterial trunk Critical aortic stenosis Corrected transposition of the great arteries Ductus arteriosus constriction Double outlet right ventricle Double inlet ventricle Ebstein s anomaly Heterotaxy syndromes Hypoplastic left heart syndrome Isolated right aortic arch Pulmonary atresia with intact ventricular septum Pulmonary atresia and ventricular septal defect Rhabdomyoma Transposition of the great arteries Tetralogy of Fallot Tricuspid atresia Tricuspid dysplasia Ventricular aneurysm Ventricular septal defect Total displayed on the nine-screen windows (Figure 1). Care was taken to have as the lowermost view (bottom right position) the transverse view of the fetal abdomen, for evaluation of the abdominal situs, and as the uppermost view (upper central position) the three vessels and trachea view with the thymus. Then, if too many windows showed intermediate non-diagnostic views (e.g. fivechamber view or incomplete view of one outflow), the slice distance was adjusted finely (increasing gradually from 1.0 mm), until all key echocardiographic views appeared in the various windows (Figure 1). The slice distance needed at the different gestational ages was recorded. The interslice distance was then regressed against gestational age and the best-fitting curve was identified. Statistical analysis was performed using the SPSS 8.0 package for Windows (SPSS, Chicago, IL, USA). The best-fit curve procedure was run and, if there were no significant differences with respect to the r 2 value, the least complex equation was chosen. RESULTS Of the various models tested, the best equation, expressing a positive correlation between interslice distance and advancing gestational age, was linear (r 2 = ; F = 953). All other tested equations, including quadratic, cubic and logarithmic ones, had similar r 2 -values but lower F-values. The distribution of best interslice distances according to gestational age is shown in Figure 2, with the corresponding means and SDs in Table 2. With respect to the simultaneous display of the central cardiovascular connections, a complete sequential segmental analysis was possible in all cases examined with TUI-STIC; some examples of complex CHD are shown in Figures 3 5. Slice distance Gestational age (weeks) Figure 2 Scatterplot showing the distribution of the interslice distance vs. gestational age (r 2 = ). The three lines indicate the mean and 95% CI

4 558 Paladini et al. Table 2 Interslice distance according to gestational age: descriptive statistics Gestational age (weeks) n Minimum Maximum Mean SD DISCUSSION The characterization of complex CHD represents one of the most difficult issues in pediatric cardiology and is even more challenging in the prenatal period, due to the anatomical and hemodynamic peculiarities of the fetal heart. Several authors have shown in the past that, despite these difficulties, a comprehensive diagnosis of CHD, with good levels of accuracy, is feasible in the fetus 4,5. However, the description of complex CHD has always been a major problem both in the diagnostic and in the teaching settings. As to the former, the sequential analysis approach adopted for the fetus 6,7 has proved helpful. However, it remains difficult to achieve immediate visual correlation of the various central connections from a Figure 3 Four-dimensional echocardiography of a 23-week fetus with double outlet right ventricle. The upper left window shows a sagittal view of the fetal thorax with the lines corresponding to the views displayed in the other windows (the dotted line corresponds to the ninth window, which is not displayed). Interslice distance = 3.0 mm. The reference dot identifies the lumen of the descending aorta in all windows. Window 3 (bottom row, right): transverse view of the abdomen, with the descending aorta (identified by the reference dot) and the stomach on the left. Window 2: inferior vena cava can be seen to the right of the aorta. Window 1: posterior four-chamber view. Window : classic four-chamber view, with good visualization of the atria, the atrioventricular plane, the ventricles and the septa. Note the reference dot within the descending aorta and the lower right pulmonary vein entering the left atrium. LV, left ventricle; pv, pulmonary veins; RA, right atrium. Window 1: modified four-chamber view, in which the ventricular septal defect (arrowhead) becomes evident. There is no sign of the aortic root. Window 2: left ventricle is no longer evident, and there is no sign of the left ventriculoarterial connection. Window 3: right outflow tract, with the main pulmonary artery (Pa) exiting from the right ventricle (RV). The arrows show the two branches of the pulmonary artery. Window 4: higher up than the pulmonary artery is another vessel, with the characteristics of the aorta (no early branching) exiting from the right ventricle. Note that the vessel is continuous with the descending aorta (marked by the reference dot). Ao, ascending aorta.

5 Role of TUI-STIC in CHD 559 Figure 4 Color Doppler imaging during systole of a 22-week fetus with tetralogy of Fallot with right aortic arch, aberrant left subclavian artery and thymic aplasia (microdeletion 22q11). The upper left window shows a sagittal view of the fetal thorax with the lines corresponding to the views displayed in the other windows (the dotted line corresponds to the ninth window, which is not displayed). Interslice distance = 2.8 mm. The reference dot identifies the lumen of the inferior vena cava (of interest in the bottom row windows). Window 3 (bottom row, right): transverse view of the abdomen, with the descending aorta (Ao) to the right of the spine, just behind the inferior vena cava (identified by the reference dot, at the confluence with the suprahepatic veins). Window 2: inferior vena cava (reference dot) is seen entering the right atrium. Window 1: posterior four-chamber view. Window : modified, anteriorized four-chamber view, with good visualization of the atria, the ventricles and the systolic ejection of blood (in red) through the aortic root. LA, left atrium; RV, right ventricle. Window 1: the malalignment ventricular septal defect (arrowhead) and the ejection of blood (in red) through the left outflow are evident. Ao, aorta; LV, left ventricle. Window 2: the initial part of the aortic arch almost reaches the right thoracic wall (a rib is seen above the Ao lettering). Window 3: right outflow tract and transverse aortic arch (AA). The main pulmonary artery (Pa) is moderately reduced in size (infundibular stenosis); the AA starts branching in front of the trachea (arrow), with the descending aorta located to the right of the trachea. Both arteries are located close to the anterior thoracic wall, because of the absence of the thymus. Window 4: the vessel branching off the aortic arch is an aberrant pre-tracheal left subclavian artery (ALSA). The arrowhead indicates the trachea. description of this type of sequential segmental approach. By showing simultaneously the central cardiovascular connections on a single panel, TUI-STIC may improve the understanding of cardiac anatomy in cases of both normal (Figure 1) and severely abnormal (Figures 3 5) cardiac connections. With this approach, all information regarding the sequential anatomy of the heart from the venous afferent vessels to the arterial outflows and arches is readily available on a single panel that includes all significant echocardiographic planes. Furthermore, the possibility of combining this approach with color Doppler enables one to convey fundamental functional information at the same time (Figures 4 and 5). Hence, we think that TUI-STIC may be considered to be one of the best tools available for optimizing the consultation of the perinatal expert with the pediatric cardiologist and the cardiac surgeon and for review by other experts in fetal cardiology. The problem of understanding the cardiovascular anatomy in cases of complex CHD is even more pronounced in the teaching setting: the perception of the spatial relationships of the heart and the great vessels poses significant difficulties to the operator with limited ultrasound experience and represents the key factor complicating the correct visualization of the outflow tracts in the screening setting. The TUI-STIC approach, with simultaneous display of all echocardiographic sections, may contribute to a better understanding of

6 560 Paladini et al. Figure 5 Color Doppler imaging during systole of a 23-week fetus with critical aortic stenosis with endocardial fibroelastosis and aortic coarctation. The upper left window shows a sagittal view of the fetal thorax with the lines corresponding to the views displayed in the other windows (the dotted line corresponds to the ninth window, which is not displayed). Interslice distance = 3.0 mm. Window 3 (bottom row, right): transverse view of the abdomen, with the descending aorta (Ao) and the inferior vena cava (IVC). Window 2: the dilatation of the left ventricle is sufficiently severe to indent the diaphragm and show up at this level. Window 1: posterior four-chamber view, with clear evidence of the huge left ventricular dilatation. The brightly echogenic endocardial rim is consistent with the presence of endocardial fibroelastosis (EFE). Window : classic four-chamber view, with good visualization of the atria and ventricles. The foramen ovale appears to be bulging within the right atrium (RA) and the inverted left-to-right shunt at this level is evident in red (arrow). LV, left ventricle. Windows 1 and 2: both windows should show the left outflow tract. However, there is only a small roundish representation of the extremely hypoplastic ascending aorta (arrowhead), with no evidence of blood flow. RV, right ventricle. Window 3: right outflow tract, showing the blood (in blue) being ejected from the right ventricle (RV) into the pulmonary artery (Pa). Window 4: three-vessels view. The image shows the hypoplastic aortic arch with reverse flow (in red), consistent with a diagnosis of severe coarctation (tubular hypoplasia) of the aortic arch (Ao). Pa, pulmonary artery and ductal arch; t, trachea. cardiovascular anatomy. However, it may be difficult for the operator to select the appropriate settings among all the possible choices offered by the ultrasound system, especially considering the changing dimensions and geometry of the heart and great vessels 10. We have tried to help solve this part of the problem by assessing the relationship between TUI-STIC interslice distance and advancing gestational age. These results may be used to choose appropriate settings at different gestational ages. It must be said that the TUI approach may also be used in a different way, especially useful for the less expert operator: rather than displaying all slices on the same panel, as in Figures 3 5, it is possible to select a singleimage format, adjust the zoom 11, and then review all transverse slices sequentially. Another innovative aspect of TUI-STIC is that it contributes significantly to the development of the automated evaluation of volumes: the final objective of volume ultrasound is a completely automated approach to the assessment of fetal anatomy, as recently underlined 12. From this standpoint, the possibility to predefine the most suitable interslice distance according to gestational age (Table 2) for the most appropriate tomographic display of all transverse echocardiographic views could lead in the near future to an automated slideshow of all these views simply by feeding into the system the key information regarding the gestational age and reference acquisition plane. The operator could then move easily from one view to the next or vice versa, without needing the additional technical experience necessary to obtain the

7 Role of TUI-STIC in CHD 561 outflow views. This, in turn, might increase significantly the currently disappointing 13,14 prenatal detection rate of CHD. The major limitation of the TUI-STIC approach to sequential anatomy is the necessity of having a volume that includes the whole of the fetal thorax, from the transverse abdominal view to the upper mediastinum. This corresponds to files that range in size from 4 to 8 megabytes for second-trimester fetuses and from 20 to 30 megabytes for third-trimester ones. Another, more important, limitation is encountered in the third trimester: in the case of advanced gestational age, the increased mineralization of the ribs and the sternum reduces significantly the number of cases in which segmental analysis can be completed with TUI-STIC. It was because of this that we had to exclude 32.2% (49/152) of our cases due to poor quality or incompleteness of the STIC volumes. However, it must be underlined that most of the 21 cases excluded for incompleteness of the volume were simple CHDs (e.g. atrioventricular septal defect), in which we had taken care to obtain a volume which best displayed the affected connection only. It is likely that, in at least some of these cases, a complete volume could have been acquired if needed. Similarly, it is important to underline that volumes, although excluded from our analysis only because of fetal movements, nonetheless contained significant anatomical information if the fetal movement occurred while the sweep of the transducer insonated anatomical structures not involved primarily in the defect (e.g. in a fetus with an atrioventricular septal defect, a movement occurring while the upper mediastinum plane was being acquired does not limit assessment of the abnormal common atrioventricular valve). Finally, a methodological limitation regards the fact that the interslice distance means and SDs (Table 2) were generated from fetuses with CHD and that it was assumed implicitly that the same distances apply also to normal hearts. While this is likely to be true, it was not demonstrated statistically in this investigation. In conclusion, we have demonstrated that the use of TUI-STIC allows the simultaneous visualization of all echocardiographic planes on a single panel, which represents the visual counterpart of the sequential segmental descriptive analysis suggested for the study of complex CHD. In addition, we have identified the best interslice distance for optimizing the visualization of echocardiographic sections with this imaging modality at different gestational ages. These data, and this kind of approach (i.e. TUI-STIC), may be employed advantageously in the characterization of complex fetal CHDs, both in the diagnostic and in the teaching settings. REFERENCES 1. Shinebourne EA, Macartney FJ, Anderson RH. Sequential chamber localization logical approach to diagnosis in congenital heart disease. Br Heart J 1976; 38: Van Praagh R. The segmental approach to diagnosis in congenital heart disease. Birth Defects Orig Artic Ser 1972; 8: Ferencz C, Rubin DJ, Loffredo AC, Magee AC (eds). Epidemiology of Congenital Heart Disease. The Baltimore Washington Infant Study Perspectives in Pediatric Cardiology, Vol. 4. Mount Kisco Futura Publishing Co: New York, 1993; Allan LD, Sharland GK, Milburn A, Lockhart SM, Groves AM, Anderson RH, Cook AC, Fagg NL. Prospective diagnosis of 1006 consecutive cases of congenital heart disease in the fetus. J Am Coll Cardiol 1994; 23: Paladini D, Russo MG, Teodoro A, Pacileo G, Capozzi G, Martinelli P, Nappi C, Calabro R. Prenatal diagnosis of congenital heart disease in the Naples area during the years the experience of a joint fetal pediatric cardiology unit. Prenat Diagn 2002; 22: Yoo SJ, Lee YH, Cho KS, Kim DY. Sequential segmental approach to fetal congenital heart disease. Cardiol Young 1999; 9: Carvalho JS, Ho SY, Shinebourne EA. Sequential segmental analysis in complex fetal cardiac abnormalities: a logical approach to diagnosis. Ultrasound Obstet Gynecol 2005; 26: Chaoui R, Hoffmann J, Heling KS. Three-dimensional (3D) and 4D color Doppler fetal echocardiography using spatio-temporal image correlation (STIC). Ultrasound Obstet Gynecol 2004; 23: Gonçalves LF, Espinoza J, Lee W, Mazor M, Romero R. Threeand four-dimensional reconstruction of the aortic and ductal arches using inversion mode: a new rendering algorithm for visualization of fluid-filled anatomical structures. Ultrasound Obstet Gynecol 2004; 24: Espinoza J, Gonçalves LF, Nien JK, Lee W, Kusanovic JP, Schoen S, Treadwell MC, Romero R. Changes in fetal cardiac geometry with gestation: implications for three-dimensional fetal echocardiography. Ultrasound Obstet Gynecol 2005; 26: Chaoui R. The four-chamber view: four reasons why it seems to fail in screening for cardiac abnormalities and suggestions to improve detection rate. Ultrasound Obstet Gynecol 2003; 22: Abuhamad A. Automated multiplanar imaging: a novel approach to ultrasonography. J Ultrasound Med 2004; 23: Todros T, Faggiano F, Chiappa E, Gaglioti P, Mitola B, Sciarrone A. Accuracy of routine ultrasonography in screening heart disease prenatally. Prenat Diagn 1997; 17: Levi S, Schnapps JP, De Havay P, Coulon R, Defort P. Endresults of routine ultrasound screening for congenital anomalies: The Belgian Multicentric Study Ultrasound Obstet Gynecol 1995; 5: