Portal de Periódicos da CAPES

Standardization of on‐screen fetal heart orientation prior to storage of spatio‐temporal image correlation (STIC) volume datasets

2007; Wiley; Volume: 29; Issue: 6 Linguagem: Inglês

10.1002/uog.4049

1469-0705

D. Paladini,

Cardiovascular Function and Risk Factors

Spatio-temporal image correlation (STIC), which allows offline analysis of dynamic volume datasets of the fetal heart, was introduced into clinical practice in 20031. Since then, its clinical role has been validated both in diagnostic2-6 and in screening7 settings. Volume datasets can be opened and processed in various ways in order to obtain the desired display of cardiovascular structures using different rendering modes (e.g. glass body, surface, inversion). However, there are a number of factors that underscore the need for standardization of storage and offline analysis of cardiac volume datasets. Spatial orientation may be difficult to determine while navigating offline. In particular, identification of left and right may be confusing after several rotations and changes of plane. In addition, the procedure needed to produce the final output, in terms of mode selection, rendering filter blend and image quality adjustment, involves numerous steps for which there are invariably several options. Hence, these steps are time-consuming and cumbersome for experts and also challenging for non-experts, who may have difficulties in identifying fetal orientation without the usual clues that realtime ultrasound provides. For these reasons, the following protocol for standardization of the orientation of the fetal heart on-screen prior to storage of dynamic volume datasets is proposed. Note that this protocol applies only to storage of STIC volume datasets and not to their acquisition, tips for which were reviewed recently by Yagel et al.8. Four-dimensional echocardiography can be performed in realtime, with matrix array transducers9, and offline, after acquisition of a dynamic volume dataset with the volumetric probes adopting the STIC algorithm1-7. This Editorial deals only with the latter approach, and the following descriptions and procedures are related to the STIC technique initially developed for the fetal heart by General Electrics (General Electrics, Kretztechnik, Zipf, Austria) and to the 4D-viewer (release 5.0) software (General Electrics) needed to open and process the volumes offline. It should be acknowledged that the nomenclature used and the software-related procedures and details may represent a limitation for the application of this standardization to other systems. However, the General Electrics Voluson platform represents the most widely available equipment worldwide for three-and four-dimensional ultrasound10, and, in addition, the described protocol for anatomical orientation may be applied to volumes obtained with other equipment. The main objective of this protocol is to make the identification of the fetal lie and sidedness on the three panels of the multiplanar imaging display as straightforward and automated as possible (Figure 1a). This will, in turn, simplify and speed up the successive steps needed to produce a rendered image of the heart. For the purposes of this Editorial, the labeling of the four panels appearing on the screen of the ultrasound system/computer after a volume has been opened is that commonly adopted and shown in Figure 1a. Gray-scale spatio-temporal image correlation (STIC) volumes of a 28-week fetus displayed in the multiplanar mode. Image (a) shows how the four panels are identified conventionally: the reference plane from which the volume is acquired is named the A-plane, the one orthogonal to the former, displayed on the top right part of the screen, is named the B-plane, and the coronal plane, shown in the bottom left panel, is named the C-plane. The remaining panel, at the bottom right, is dedicated to the rendered image and is therefore named the R-panel. The overlay in (b) shows how, if the fetus is in vertex presentation, in the A-plane, the left ventricle is positioned on the left part of the image, and in the B-plane, the head of the fetus is on the left and the breech is on the right. The overlay in (c) shows how, if the fetus is in breech presentation, on the A-plane, the left ventricle lies on the right side of the image, and in the B-plane, the head of the fetus is on the right. To transform the volume in order to store it as if the fetus were in vertex presentation, it is sufficient to rotate by 180° the volume along the y-axis, by dragging the corresponding cursor below the image (arrow) to the end of the run of either side of the scroll bar. LV, left ventricle; RA, right atrium; RV, right ventricle. For the purposes of this standardization protocol, an apical four-chamber view is defined as an axial view of the fetal thorax at the level of the heart, wherein the angle between the insonating beam and the cardiac axis, identified with respect to the interventricular septum, is in the range of 0 ± 45°. In this view, no rotation should be applied to the z-axis, in order not to alter the normal aspect of the image in terms of ultrasound-wave reflection and refraction. The heart should be displayed with its left side on the left part of the A-panel (Figure 1a), so that on the B-panel the fetal head will lie on the left and the fetal breech on the right. This position corresponds to that of a fetus in vertex presentation (Figure 1b). If the fetus is lying in breech presentation, which causes the left ventricle to appear on the right side of the image with the fetal head on the right of the B-panel (Figure 1c), it is necessary to rotate by 180° the image on the y-axis in order to flip it horizontally. In this way, the final appearance will be that shown in Figure 1b. This 180° rotation can be achieved by dragging the cursor of the y-axis to the end of the run on either side of the scroll bar (Figure 1c). Use of this view will ensure that the spatial orientation of the fetus will always be the same, with the head on the left side of the B-panel. This facilitates the selection of the correct option in the region of interest (ROI) direction menu (which includes six possible choices; Figure 2) when the rendering mode is activated. To further standardize the procedure, the ROI direction options can be labeled arbitrarily from 1 to 6 as shown in Figure 2. Thus, for renderings that require a craniocaudal direction of the ROI, including the glass body and inversion renderings of the crossover of the vessels and of the upper mediastinal view (Figure 3), and for the glass body, inversion and surface rendering of the four-chamber view (Figure 4), ROI Option 3 should be selected. However, no change in the ROI direction (i.e. ROI Option 1) is required for the en-face view of the atrioventricular and ventriculoarterial valves (Figure 5). Alternatively, some operators prefer to obtain the same view from the ventricular side, especially if the leaflets of the atrioventricular valves need be demonstrated; in this case, ROI Option 2 is required. When the render option is activated on the 4D-viewer software, the operator should change the direction of the region of interest (ROI) according to the type of rendered image he/she wants to produce. When the ‘ROI direction’ submenu is activated under the ‘Settings’ drop-down menu, the operator is presented with six possible options, corresponding to each of the possible directions of the ROI on the three orthogonal planes (arrowheads). To standardize the procedure, the labeling displayed here from 1 to 6 (top left to bottom right) is proposed as standard. Glass body (a) and inversion (b) rendering demonstrating the normal crossover of the great arteries in a 28-week fetus. Both these renderings require the region of interest direction Option 3 to be activated. Arrowheads indicate right and left pulmonary arteries. Ao, ascending aorta; LV, left ventricle; Pa, main pulmonary artery; RV, right ventricle. Surface (a) and inversion (b) rendering of a normal four-chamber view (28 and 23 weeks of gestation, respectively). As in Figure 3, these renderings require region of interest (ROI) direction Option 3 to be activated, but require ROI thickness that is less pronounced, because only the four-chamber view need be included. The arrow indicates the inferior vena cava entering the right atrium, and arrowheads indicate the descending thoracic aorta. LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle. En-face view of the four cardiac valves. This rendering requires region of interest direction Option 1 to be activated. In the surface-rendered image (a), the leaflets of the tricuspid (TV and arrows) and the mitral (MV and arrowheads) valves are demonstrated (22 weeks of gestation); the glass body rendering (b) demonstrates the normal spatial relationships of the two atrioventricular and the two ventriculoarterial valves, with the aorta wedged in between the two atrioventricular valves (28 weeks of gestation). Ao, aortic outflow; M, flow across the mitral valve; Pa, pulmonary outflow; T, flow across the tricuspid valve. It should be emphasized that the thickness of the ROI needed to display the various views of the anatomical structures mentioned above will vary according to the endpoint of the rendering: thick slices including the whole of the heart and great vessels will be required to demonstrate the crossover of the vessels; relatively thin slices will be required to display the four-chamber view in surface-rendering/inversion modes; thin or very thin slices will be required to display the en-face view of the four cardiac valves. Since the apex of the heart is rotated to the left (levocardia: the normal cardiac axis is 45 ± 20°), a transverse four-chamber view is usually achieved with insonation from the right side of the trunk, so that the left ventricle will be the one farthest from the transducer. For the purposes of this standardization protocol, a four-chamber view is defined as being transverse if the angle between the insonating beam and the cardiac axis, identified with respect to the interventricular septum, is in the range of 90 ± 45°. In this view, as for the apical four-chamber view, no rotation should be applied to the z-axis, in order not to alter the normal aspect of the image in terms of ultrasound-wave reflection and refraction. The heart should be displayed with the left ventricle on the lower part of the A-panel, with the apex pointing to the left (Figure 6), so that on the B-panel the fetal head will lie on the left and the fetal breech on the right, as for the apical four-chamber view. This position corresponds to that of a fetus in vertex presentation (Figure 1b). If the fetus is lying in breech presentation, which causes the cardiac apex to appear on the right side of the image (Figure 7), it is necessary to rotate by 180° the image on the y-axis in order to flip the image horizontally. In this way, the final appearance will be that shown in Figure 6. This 180° rotation can be obtained by dragging the cursor of the y-axis to the end of the run on either side of the scroll bar (Figure 1c). Transverse four-chamber view: gray-scale spatio-temporal image correlation (STIC) volume of a 24-week fetus displayed in the multiplanar mode. In this case, the fetus was in vertex presentation and, therefore, in the A-plane, the left ventricle is positioned on the far side of the screen relative to the transducer and the cardiac apex points to the left of the image (arrowhead); in the B-plane, the head of the fetus is on the left and the breech is on the right. Note the stomach on the right of the B-plane image, below the diaphragm. LV, left ventricle; RA, right atrium. Transverse four-chamber view. If the fetus is in breech presentation, in the A-plane, the left ventricle will always lie on the far side of the image, but the cardiac apex will point to the right of the image (arrowhead); in the B-plane, the head of the fetus will be on the right. To transform the volume in order to store it as if the fetus were in vertex presentation, it is sufficient to rotate by 180° the volume along the y-axis as shown for the apical four-chamber view in Figure 1c: this is accomplished by simply dragging the corresponding cursor below the image (Figure 1c, arrow) to the end of the run at either side of the scroll bar. LV, left ventricle; RA, right atrium. It should be borne in mind that, on gray-scale imaging, volumes that have as their reference plane the transverse four-chamber view are diagnostically as good as are those that have the apical four-chamber view as their reference plane. However, renderings that rely on the color Doppler signal to highlight cardiac chambers and valves, such as the en-face view of the four cardiac valves, cannot be achieved with this approach. The reason for this relates to the physics of the Doppler principle and the fact that ventricular inflow will be at 90° to the insonating beam and thus not amenable to fast Fourier transformation. An example to demonstrate the application of this standardization protocol is illustrated in Figure 8. In a case of atrioventricular discordance, such as occurs in corrected transposition of the great arteries, the assignment of the cardiac chambers may be difficult if the operator is not certain of the situs. Moreover, corrected transposition can be associated with dextrocardia11 and this would make the identification of right and left sides of the fetus even more problematic. Following the proposed standardization protocol, in the A-panel, the right side of the thorax is the one closest to the transducer. In Figure 8, it can be seen that the morphological right ventricle (round shape and tricuspid valve with chordae tendinae attaching at the apex of the ventricle), connected with the left atrium, is positioned on the left, in the far ultrasound field; conversely, the morphological left ventricle (forming the apex of the heart, with the chordae tendinae of the mitral valve attaching on the lateral myocardial wall) is on the right, in the near ultrasound field. If the side of the fetus is not known and the operator opening the volume is not an expert, he/she may become confused and identify erroneously the right side with the morphological right ventricle, overlooking the atrioventricular discordance. However, the standardization protocol having been applied prior to storage would automatically identify the sidedness of the fetus and reduce the chances of error. Transverse four-chamber view from a spatio-temporal image correlation (STIC) volume of a 28-week fetus with corrected transposition of the great arteries, which is characterized by atrioventricular discordance. The round shape of the morphological right ventricle (mRV) and the tricuspid valve with chordae tendinae attaching at the ventricular apex (arrow) are visible on the left, in the far ultrasound field; conversely, the morphological left ventricle (mLV), forming the apex of the heart, with the chordae tendinae of the mitral valve attaching on the lateral myocardial wall (arrowhead), is on the right, in the near ultrasound field. LA, left atrium; LT, left hemithorax; RA, right atrium; RT, right hemithorax. It should be stressed that the stomach is not a reliable indicator of sidedness, because it can be in the right hemiabdomen if intestinal malrotation or situs inversus are associated anomalies. From the point of view of the expert, it is of fundamental importance to be able to assign correctly the situs. Knowing that the standardization protocol has been applied and, hence, that the appearance of the cardiac chambers follows the criteria described above, is vital to a reliable sequential cardiac anatomical analysis12. An important contribution from the ultrasound equipment manufacturers would be to add to their equipment commands an icon which might be attached to STIC volumes to which this standardization protocol (or any modification thereof) has been applied prior to storage. In this way, anyone opening the same volume will recognize easily and with confidence the fetal lie and sidedness. In summary, in this Editorial, I have proposed standardization of orientation of the fetal heart on the ultrasound screen prior to STIC volume storage. My hope is that this will form the basis of a discussion and that readers' comments and suggestions for modifications of the protocol described here will be collated in order to reach a consensus. In this way, life will be made much easier for everybody who is confronted with the thrilling but challenging task of handling a four-dimensional volume dataset of the fetal heart acquired with the STIC technique.