Portal de Periódicos da CAPES

How do we do it? Practical advice on imaging‐based techniques and investigations

2006; Wiley; Volume: 27; Issue: 3 Linguagem: Inglês

10.1002/uog.2724

1469-0705

Luís F. Gonçalves, W. Lee, Jimmy Espinoza, Roberto Romero,

Congenital Diaphragmatic Hernia Studies

Three-dimensional (3D) and four-dimensional (4D) ultrasound have been proposed to be valuable tools for the examination of the fetal heart1-52. The ultimate goal of 3D and 4D ultrasound is to improve the detection rates of congenital heart disease53-63 by decreasing the dependency on operator skills required by two-dimensional (2D) ultrasound64, 65. This is important since congenital heart disease is the leading cause of death among infants with congenital anomalies66, and prenatal diagnosis is associated with decreased neonatal morbidity and mortality rates67-70. Several techniques for prenatal examination of the fetal heart by 3D/4D ultrasound have been proposed using a variety of technologies, including free-hand 3D ultrasound with and without position sensors1, 3, 7-9, 12, 30, 3D ultrasound with automated mechanical acquisition4, 5, 7, 16, 26, 4D ultrasound with a variety of gating algorithms (temporal Fourier analysis of heart motion3, M-Mode1, Doppler11, 17-19, 39, 49, spatio-temporal image correlation (STIC)27, 28, 34), as well as real-time 4D ultrasound with 2D matrix-array transducers10, 15, 24, 31, 46, 52. In this article, we describe a practical approach for the examination of the fetal heart using 4D ultrasound with STIC. STIC is a technique that allows acquisition of a fetal heart volume and visualization of cardiac structures as a 4D cine sequence containing information of one full cardiac cycle27, 28, 34. The principles utilized by the STIC algorithm to synchronize spatial and temporal information (i.e. the 3D images plus motion) in volume datasets of the fetal heart are similar to the non-ECG motion Fourier analysis gating method proposed by Nelson et al.3 in 1996, and have been described in detail elsewhere27, 28, 34. The acquisition of high-quality volume datasets is crucial for the successful examination of the fetal heart by 4D ultrasound. Therefore, the 2D images that originate the 4D volume datasets must be optimized prior to acquisition to maximize the frame rate and improve contrast resolution. Likewise, color and power Doppler settings must be optimized prior to acquisition. Once the 2D image is optimized, a region of interest (ROI) encompassing the heart and any other structures of interest (e.g. stomach, spine) is selected by the user and volume acquisition is initiated (Figure 1, S1). Below we describe a few tips that we have found to be useful during volume acquisition: Two-dimensional image of the four-chamber view, illustrating the placement of the region of interest (ROI) prior to acquisition of the volume dataset. The ROI defines the width and height of the volume dataset (x- and y-planes). LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle. If the examiner is interested in the evaluation of the four-chamber, five-chamber, three-vessel, and three vessels and trachea views, optimal volume datasets are acquired using transverse sweeps through the fetal thorax. Instead, if the examiner wants to review the aortic and ductal arches, or the venous return to the heart, high-quality volume datasets are best acquired using sagittal sweeps through the fetal thorax. The ideal fetal position for volume acquisition is one in which the fetus is lying on its back (i.e. with the spine positioned at 6 o'clock). However, high-quality volume datasets can also be obtained when the fetal spine is up, provided that it is not positioned between 11 and 1 o'clock. Visualization of fetal heart structures in volume datasets acquired with the spine positioned between 11 and 1 o'clock are frequently compromised by acoustic shadowing from the spine and ribs. The ROI determines the width and height of the volume dataset (x- and y-planes). Although a large ROI may be selected to include all structures that surround the heart, the reader is reminded that wide ROIs are associated with low frame rates. This may have a negative impact on the quality of the volume dataset, especially if color or power Doppler is utilized during acquisition. Thus, the selection of an ROI as narrow as possible maximizes the frame rate during acquisition and improves the temporal resolution of the volume dataset. The acquisition angle determines acquisition depth (i.e. the amount of information that is acquired in the z- or azimuth plane). For second-trimester fetuses, acquisition angles of between 20° and 25° are usually sufficient to include the stomach, the heart, and the great vessels in the volume dataset. For larger fetuses, acquisition angles of between 35° and 40° may be necessary. The acquisition time can be selected by the user and usually ranges from 7.5 to 15 s. The acquisition time determines the speed at which the transducer sweeps the ROI. The longer the acquisition time, the higher the spatial resolution of the volume dataset. Ideally, volume datasets should be acquired using the longest possible acquisition time. However, the longer the sweep takes, the higher the likelihood that artifacts related to fetal motion may be present in the volume dataset. Thus, for highly active fetuses, there may be no alternative but to scan with the fastest possible acquisition time, at the expense of optimal spatial resolution. The most basic approach for the examination of a volume dataset of the fetal heart acquired with STIC is to simply scroll through the volume from top to bottom along the original plane of acquisition. This is illustrated in Figure 2 ( S2) and has recently been facilitated by the development of automatic tomographic displays (Tomographic Ultrasound Imaging, GE Healthcare, Milwaukee, WI, USA) which have been incorporated into the ultrasound equipment (VOLUSON 730 Expert, GE Healthcare) and offline analysis software (4DView version 5.0, GE Healthcare) (Figure 3, S3). This approach allows the examiner to determine quickly the relationship between the apex of the heart and stomach, and to evaluate the four-chamber view, and, in most cases, the five-chamber and three-vessel views. Standard planes of section that can be obtained by scrolling from the top (a) to the bottom (d) of the volume dataset. (a) Three vessels and trachea view. (b) Five-chamber view. (c) Four-chamber view. (d) Transverse section through the upper portion of the abdomen. DAo, descending aorta; LV, left ventricle; LVOT, left ventricular outflow tract; PA, pulmonary artery; RV, right ventricle; S, stomach; SVC, superior vena cava; TAo, transverse aortic arch. Tomographic ultrasound imaging of a normal volume dataset of the fetal heart, acquired using transverse sweeps through the fetal thorax. Eight cross-sections from the upper mediastinum to the upper abdomen are displayed. (a) The 'overview image': the parallel lines indicate the exact position of the planes of section shown in the eight subsequent images (b–i). (c) Three-vessel view. (e) Five-chamber view. (f) Four-chamber view. (i) A transverse section through the upper abdomen. The same image, with motion, is shown in Videoclip S3. LA, left atrium; LV, left ventricle; LVOT, left ventricular outflow tract; PA, pulmonary artery; RA, right atrium; RV, right ventricle; S, stomach; SVC, superior vena cava; TAo, transverse aortic arch. In 2003, we proposed, and subsequently validated, a technique to demonstrate consistently the left and right outflow tracts in volume datasets acquired with STIC28, 71. The technique consists of a series of systematic manipulations of the volume dataset, resulting in the visualization of the left outflow tract in panel A and the short-axis view of the pulmonary artery in panel B (Figure 4, S4). Figure 5 illustrates how to use this technique to diagnose transposition of the great arteries (Figure 5, S5). Technique to visualize systematically the left and right outflow tracts in volume datasets acquired with spatio-temporal image correlation (STIC). Only panels A (original plane of acquisition, transverse) and B (sagittal plane) are displayed. (a) The first step in this technique consists of making sure that the left side of the heart is located on the left side of the image, and the right side of the heart is on the right side of the image; if necessary, rotate the volume around the y-axis until this is achieved. The reference dot is then positioned in the crux of the heart, on both the transverse and the sagittal planes. (b) The volume dataset is rotated around the z-axis until a perfect apical four-chamber view is obtained. (c) The volume is then rotated around the z-axis counter-clockwise until the angle between the apex and the transducer is approximately 30–40°. (d) The crucial step is to position the reference dot in the interventricular septum, midway between the crux of the heart and the apex; this will anchor the three orthogonal planes for the next rotation movement, which will display the left ventricular outflow tract. (e) The volume dataset is now rotated around the y-axis; this will 'open up' the continuity between the interventricular septum and the anterior wall of the aorta; the anterior leaflet of the mitral valve is seen in continuity with the posterior wall of the aorta. (f) Once the reference dot is moved above the aortic valve, the short-axis view of the right ventricular outflow tract is displayed on the sagittal plane. These steps are illustrated dynamically in Videoclip S4. IVS, interventricular septum; LVOT, left ventricular outflow tract; RV, right ventricle. Visualization of transposition of the great arteries using the technique described in Figure 4. (a) Apical four-chamber view. (b) The volume dataset is rotated 30–40° around the z-axis. (c) The reference dot is anchored in the interventricular septum. (d) Upon rotation of the volume dataset around the y-axis, two parallel vessels exiting the ventricles are visualized. The pulmonary artery (PA—note the bifurcation) leaves the left ventricle (LV), whereas the aorta (Ao) leaves the right ventricle (RV). IVS, interventricular septum. Visualization of the aortic and ductal arches is best accomplished in volume datasets acquired using sagittal sweeps through the fetal chest. The original technique describing how to visualize systematically the aortic and ductal arches was published by Bega et al.16 in 2001, for the examination of 3D volume datasets (Figure 6, S6). 3D multiplanar slicing of the aortic and ductal arches. (a) 3D multiplanar image of the fetal thorax in the original acquisition plane (sagittal). (b) The reference dot was manipulated in the right upper panel and moved to the center of the aorta (white arrow). The resulting image in the left upper panel was a sagittal view of the aorta. Minor adjustment of this image around the y-axis was required to demonstrate the aortic arch. (c) To demonstrate the ductal arch, the upper right panel image was simply rotated counter-clockwise around the z-axis (curved arrow). Reproduced from Gonçalves LF, Lee W, Chaiworapongsa T, Espinoza J, Schoen ML, Falkensammer P, Treadwell M, Romero R. Four-dimensional ultrasonography of the fetal heart with spatiotemporal image correlation. Am J Obstet Gynecol 2003; 189: 1792–1802. Copyright © 2003 Mosby, Inc. In 2004, DeVore et al.41 proposed a technique to display any structure of interest during the examination of volume datasets of the fetal heart, simply by positioning the reference dot in the center of the structure of interest and 'spinning' the volume dataset around the y-axis until the structure is visualized in its entirety. For a detailed description of this valuable technique, we refer the reader to the original article41. Rendering techniques can be used to visualize any intracardiac structure. These techniques provide a depth perspective to the structure being examined, and can be used to optimize the contrast of the myocardial borders, septa and valves, or to obtain realistic 4D images of particular structures of interest. In Figure 7 ( S7), 'thick-slice' rendering was applied to the atrioventricular valves to visualize the leaflets en face, as if the examiner was observing them from the ventricular chambers28, 40. Figure 8 ( S8) demonstrates the use of 'thick-slice' rendering with inversion mode to emphasize the abnormal insertion of the tricuspid valve in a case of Ebstein anomaly48. In Figure 9 ( S9) we provide another example of thick-slice rendering to visualize the overriding aorta in a case of tetralogy of Fallot associated with absent pulmonary valve syndrome. Figure 10 ( S10) shows the pulmonary artery, which is constricted at the level of the pulmonary valves, and the poststenotic dilatation. This image also shows, as a result of the depth perspective provided by rendering, a cross-section of a dilated left pulmonary artery, a common finding in absent pulmonary valve syndrome. 'Thick-slice' rendering of the atrioventricular valves of a normal fetus. The key to obtaining the rendered image (lower right corner in a and b) is the position and size of the region of interest, encompassing only the atrioventricular valves. The green line indicates the direction of view, which is the direction that the computer software will use to convert the voxels along the projection path into pixel information to be displayed on the two-dimensional screen. In this case, the position of the green line indicates that the atrioventricular valves are being visualized from the ventricular chambers. This volume dataset was rendered using a combination (mix) of 60% gradient light and 40% surface mode. LV, left ventricle; MV, mitral valve; RV, right ventricle; TV, tricuspid valve. 'Thick-slice' rendering of the atrioventricular valves using inversion mode in a fetus with Ebstein anomaly. This mode provides great contrast for the visualization of the myocardium and atrioventricular valves. Note the abnormal apical insertion of the tricuspid valve and the small right ventricle. IVS, interventricular septum; LV, left ventricle; MV, mitral valve; RV, right ventricle; TV, tricuspid valve. 'Thick-slice' rendered image of the aorta (Ao) overriding the interventricular septum (IVS) in a case of absent pulmonary valve syndrome associated with tetralogy of Fallot. The algorithm used for rendering was the 'gradient light' mode. LV, left ventricle; RV, right ventricle. 'Thick-slice' rendered image of the right ventricle (RV) and pulmonary artery (PA) in a case of absent pulmonary valve syndrome associated with tetralogy of Fallot. The stenotic pulmonary valve annulus (PVA), the poststenotic dilatation of the pulmonary artery and a cross-section of the dilated annulus of the left pulmonary artery (LPA) are demonstrated in a single image. The algorithm used for rendering was the 'gradient light' mode. IVS, interventricular septum; LV, left ventricle. Perhaps the most impressive and unique images that can be obtained with 3D/4D techniques for fetal cardiac imaging are those that allow visualization of the 3D structure and spatial relationships of the great vessels and venous return to the heart. These images can be generated by acquiring volume datasets with color Doppler23, 25, 40, 45, power Doppler40, 45, 72 or B-flow imaging48, as well as by rendering gray-scale volume datasets with inversion mode3, 43, 47, 48. Crisscrossing of the pulmonary artery over the ascending aorta as these vessels leave the ventricular chambers is best visualized in volume datasets acquired using transverse sweeps through the fetal thorax. Sagittal acquisitions are preferred when the objective is to visualize the aortic and ductal arches. Figure 11 ( S11) shows the technique that we use to visualize the crisscrossing of the outflow tracts as they leave the ventricular chambers. As described in the previous paragraph, these volume datasets are acquired using the four-chamber view as the starting point. For the sake of reproducibility, we reorient the position of the heart on the screen, if necessary, by rotating the volume dataset around the y- and z-axes until the four-chamber view is in the apical position, and the left side of the heart is displayed on the left side of the screen. Next, the rendering box is selected and adjusted in panel B to include the whole heart within the ROI to be rendered (from its base touching the diaphragm, to the great vessels close to the neck). In order to visualize the great vessels leaving the heart, the user must set the direction of view to project the heart from the great vessels towards its base. In our example, the direction of view was set from left to right, using panel B as a reference. The resulting rendered image is shown in panel D. Figure 12 shows examples of the pulmonary artery crisscrossing over the aorta in normal volume datasets acquired with color Doppler (Figure 12a), power Doppler (Figure 12b), inversion mode (Figure 12c) and B-flow imaging (Figure 12d and S12). Technique to obtain rendered images of the outflow tracts using color Doppler. The rendering box is adjusted to encompass the heart and great vessels. The direction of view (green line) is set to project the rendered image from anterior (pulmonary artery) to posterior (aorta and ventricular chambers). Videoclip S11 shows, in detail, the steps required to obtain rendered views of the outflow tracts. The same technique can be applied to volume datasets acquired with power Doppler and B-flow imaging, as well as to volume datasets acquired with B-mode imaging but rendered using inversion mode. (a) Transverse plane showing the four-chamber view. (b) Sagittal plane orthogonal to the transverse plane. (c) Coronal plane. (d) Rendered view of the outflow tracts. Ao, aorta; LV, left ventricle; PA, pulmonary artery; RV, right ventricle. Crisscrossing of the outflow tracts as they exit the ventricular chambers. The pulmonary artery always crosses in front of the aorta. The rendered images were obtained with the technique described in Figure 11 and are explained, in detail, in Videoclip S11. The volume datasets were acquired with (a) color Doppler, (b) power Doppler, (c) gray scale (then rendered with inversion mode), and (d) B-flow. Ao, aorta; PA, pulmonary artery. Inversion of gray-scale voxels to visualize blood pools from cardiac structures was originally proposed by Nelson et al.3 in 1996. Recently this principle has been incorporated into commercially available ultrasound equipment and is known as inversion mode. With inversion mode, anechoic structures such as the heart chambers, vessel lumen, stomach, gallbladder, renal pelvis and bladder appear echogenic in the rendered images, whereas structures that are normally echogenic before gray-scale inversion (e.g. bones) appear anechoic. Postprocessing adjustments are performed as necessary, including gamma-curve correction to optimize contrast resolution, and gray-scale threshold and transparency correction to improve image quality. The technique allows examiners to obtain 4D rendered images of cardiovascular structures from volume datasets acquired with gray scale only, without the need for color Doppler, power Doppler or B-flow imaging. In S13, we describe a technique to obtain 'digital casts' of the aortic and ductal arches from volume datasets acquired using sagittal sweeps through the fetal chest (Figure 13). 'Digital casts' of the aortic and ductal arches obtained with inversion mode. AAo, ascending aorta; D, diaphragm; DA, ductus arteriosus; DAo, descending aorta; E, esophagus; G, gallbladder; IS, intervertebral space; IVC, inferior vena cava; LiV, liver vasculature; PA, pulmonary artery; PV, portal vein; S, stomach; SVC, superior vena cava. Reproduced from Gonçalves et al.43. B-flow technology digitally enhances weak blood reflector signals from vessels, at the same time suppressing strong signals from the surrounding tissues73-76. Because this technology does not use Doppler methods to display blood flow, it is angle-independent and does not interfere with the frame rate as much as color or power Doppler75, 77-79. Because of its high sensitivity and angle independence to blood flow, B-flow is potentially advantageous over color or power Doppler imaging for the visualization of the great vessels and venous return to the heart48, 76. In Figure 14 ( S14), the aortic and ductal arches of a normal fetus have been rendered using a volume dataset acquired with B-flow imaging and the gradient light algorithm. The principles utilized to obtain these images are the same as those described for Figures 11, 12 and 13. Rendered image of the aortic and ductal arches obtained from a volume dataset acquired with B-flow imaging, using a similar acquisition and rendering technique to that illustrated in Videoclip S13. Ao arch, aortic arch; DA, ductus arteriosus; DAo, descending aorta; IVC, inferior vena cava; PA, pulmonary artery. We have described a practical approach for the examination of the fetal heart by 3D/4D ultrasound with STIC. This set of tools for fetal cardiac examination allows: (1) the possibility of navigating through the volume dataset and examining the fetal heart in the absence of the patient; (2) the use of techniques to visualize systematically the outflow tracts in volume datasets acquired using the four-chamber view image as the starting 2D image; (3) examination of the fetal heart using a tomographic approach similar to that used to read computerized tomography and magnetic resonance imaging examinations; (4) 3D and 4D rendering of cardiovascular structures to visualize the relationships, size and course of the outflow tracts in normal fetuses and those with congenital heart disease. 3D and 4D rendering of the great vessels in particular has only been possible in the past during postmortem examinations, by injection of silicon rubber to produce pathological casts of the cardiovascular system80, 81. The following videoclips are available from Wiley Online Library: http://onlinelibrary.wiley.com/doi/10.1002/uog.2724/suppinfo (restricted access) S1 Two-dimensional videoclip of the four-chamber view, illustrating the placement of the region of interest (ROI) and acquisition of the volume dataset with spatio-temporal image correlation (STIC). The ROI defines the width and the height of the volume dataset (x- and y-planes). S2 Standard planes of section that can be obtained by scrolling from the top to the bottom of the volume dataset. S3 Tomographic ultrasound imaging of a normal volume dataset of the fetal heart, acquired using transverse sweeps through the fetal thorax. S4 Demonstration of the technique to visualize systematically the outflow tracts in volume datasets acquired using transverse sweeps through the fetal thorax with spatio-temporal image correlation (STIC). S5 Demonstration of transposition of the great arteries using the technique described in Figure 4 and Videoclip S4. S6 Demonstration of a technique to visualize the aortic and ductal arches in 4D volume datasets acquired using sagittal sweeps through the fetal chest (described originally by Bega et al.16, for volume datasets acquired by 3D ultrasound). S7 'Thick-slice' rendering of the atrioventricular valves of a normal fetus. The key to obtaining the rendered image (lower right corner) is the position and size of the region of interest, encompassing only the atrioventricular valves. The green line indicates the direction of view, which is the direction that the computer software will use to convert the voxels along the projection path into pixel information to be displayed on the two-dimensional screen. In this case, the position of the green line indicates that the atrioventricular valves are being visualized from the ventricular chambers. This volume dataset was rendered using a combination (mix) of 60% gradient light and 40% surface mode. S8 'Thick-slice' rendering of the atrioventricular valves using inversion mode in a fetus with Ebstein anomaly. This mode provides great contrast for the visualization of the myocardium and atrioventricular valves. Note the abnormal apical insertion of the tricuspid valve. S9 'Thick-slice' rendering in a case of absent pulmonary valve syndrome associated with tetralogy of Fallot, demonstrating the overriding aorta. The algorithm used for rendering was the 'gradient light' mode. S10 'Thick-slice' rendering of the right ventricle and pulmonary artery in a case of absent pulmonary valve syndrome associated with tetralogy of Fallot. The stenotic pulmonary valve annulus, the poststenotic dilatation of the pulmonary artery and a cross-section of the dilated annulus of the left pulmonary artery are demonstrated in a single image. The algorithm used for rendering was the 'gradient light' mode. S11 Technique to obtain rendered images of the outflow tracts using color Doppler. The same technique can be applied to volume datasets acquired with power Doppler and B-flow imaging, as well as for volume datasets acquired with B-mode imaging but rendered using inversion mode. Ao, aorta; LV, left ventricle; PA, pulmonary artery; RV, right ventricle. S12 Crisscrossing of the outflow tracts in a volume dataset acquired with B-flow imaging. S13 Technique to render the aortic and ductal arches using volume datasets acquired with gray-scale imaging only and rendered with inversion mode. S14 Four-dimensional visualization of the aortic and ductal arches in a volume dataset acquired with B-flow imaging. This research was supported by the Intramural Research Program of the National Institute of Child Health and Human Development, NIH, DHHS. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.