Three-Dimensional Transesophageal Echocardiography of Atrial Septal Defect: A Qualitative and Quantitative Anatomic Study


Real-time three-dimensional (3D) transesophageal echocardiography (TEE) was used to analyze atrial septal defect (ASD) with 4 goals: (1) to determine feasibility, (2) to analyze the accuracy of qualitative and quantitative data, (3) to assess strengths and weaknesses of the available modes of 3D TEE, and (4) to provide 3D transesophageal echocardiographic reference images.


Sixty-five patients with ASDs (age, 5–64 years; weight, 20–114 kg; body surface area, 0.8–2.4 m 2 ) underwent 3D TEE during catheter intervention or surgery. Three-dimensional transesophageal echocardiographic formats included live 3D, 3D zoom, and full-volume 3D modes. Qualitative and quantitative analysis of the 3D data was compared with two-dimensional echocardiographic data and intraoperative inspection.


Diagnostic-quality 3D TEE was successfully performed in all 65 patients. Fifty had secundum ASDs and 15 had other ASD types (seven sinus venosus, six primum, one common atrium, and one coronary sinus ASD). ASD type and location were correctly diagnosed in all patients. ASD shape and orientation were confirmed in 21 patients at surgery. Quantitative analysis of ASDs successfully demonstrated rims and changes in dimensions from systole to diastole. Live 3D mode had the highest volume rate, the best transgastric views, and the best views during device deployment but was limited by small sector size. Three-dimensional zoom mode allowed precropped live 3D images but was limited by slow volume rate. Full-volume mode had the best views of large defects and surrounding anatomy. However, it was limited by stitch artifact and required postacquisition cropping.


Three-dimensional TEE is feasible and accurate. Each of the 3D transesophageal echocardiographic modalities has strengths and limitations.

Atrial septal defect (ASD) is the fourth most common congenital cardiac anomaly, with an incidence of approximately 1 per 1,000 live births. There are five types of ASDs, which in decreasing order of frequency are secundum ASD (ASD2), primum ASD, sinus venosus ASD, common atrium, and coronary sinus ASD. Because of the untoward effects of a chronic left-to-right shunt, an ASD of significant size requires closure by device or surgery. Most ASD2s are amenable to transcatheter device closure, but the other types of ASDs are closed surgically.

Transesophageal echocardiography (TEE) is an integral part of the assessment of patients with ASD2s undergoing transcatheter and operative treatment and in those with nondiagnostic transthoracic imaging. The usefulness and accuracy of two-dimensional (2D) TEE during ASD device closure has been established. Furthermore, initial reports of the use of three-dimensional (3D) transthoracic echocardiography and 3D TEE for guidance of ASD device closure have demonstrated its usefulness.

The development of 3D transesophageal echocardiographic capabilities along with dedicated and rapid 3D software provides a new and improved tool to analyze the 3D anatomy of an ASD and its dynamic changes. We used matrix-array real-time live 3D TEE to image 65 patients with ASDs to (1) determine its feasibility, (2) analyze the accuracy of its qualitative and quantitative data, (3) assess the strengths and weaknesses of the three available modes of 3D TEE, and (4) provide a collection of reference images.


Sixty-five consecutive patients with ASDs underwent 2D TEE for clinical indications. All studies were performed with the patient intubated and under general anesthesia or sedated with midazolam and fentanyl. Three-dimensional TEE was also performed using live, zoom, and full-volume modes with commercially available ultrasound equipment (iE33 ultrasound machine, X7-2t 3D transesophageal echocardiographic transducer, image processing and cropping software, and 3DQ quantification software; Philips Medical Systems, Andover, MA). Patients were enrolled prospectively and consecutively. The only exclusion criterion was body weight < 20 kg.

Three-Dimensional Transesophageal Echocardiographic Protocol

Three-dimensional images of the ASD were acquired from the upper esophagus basal transverse view at 0° and the basal short-axis view at 30° to 60°. A bicaval view was obtained at the midesophageal level with 90° to 120° array orientation. The four-chamber view was obtained from the midesophageal level at 0° to 20° array orientation. A sagittal bicaval view was obtained from the deep transgastric position with a transducer array orientation of 100° to 120°. The posterior view was obtained by two methods: (1) From a transgastric sagittal bicaval view using live 3D mode, the standard image was rotated around the anatomic vertical axis by 90° to 110° to demonstrate the posterior view. (2) From the midesophageal long-axis view with 110° transducer array, the bicaval view was acquired using 3D zoom mode. The superior vena cava (SVC) was rotated into a vertical position. Next, the image was rotated 180° around the vertical axis so that the posterior aspect of the heart was facing the viewer. Last, the image was cropped from the anatomic posterior to anterior direction.

Additional or modified images were performed in some cases at the discretion of the echocardiographer to provide optimal delineation of the ASD and surrounding structures. Images were acquired from all of the above transducer positions using all three different 3D echocardiographic modes, including live 3D, 3D zoom, and full-volume 3D imaging. Specific anatomic features analyzed are described below. Also, in patients with ASD2s, real-time monitoring of device deployment and assessment of position and configuration were performed.

Three-Dimensional Echocardiographic Modes

For anatomic assessment, we used three different 3D echo modes: live 3D, 3D zoom, and full-volume 3D imaging. These three modes are all available for both 3D transthoracic echocardiography and 3D TEE. In this study, we focused on the 3D transesophageal echocardiographic application of the live 3D, 3D zoom, and 3D full-volume modes.

Live 3D mode involves real-time live imaging with a wedge-shaped 3D sector 15° to 20° thick and widest at its most distal extent. The sector volume expands in height and width as it proceeds farther away from the transducer. This mode requires real-time continuous steering, analogous to 2D echocardiographic transducer manipulation. The sonographer usually places the structure of interest in the far field, where the 3D sector is largest, to achieve the maximum 3D echocardiographic capabilities. The volume rate is fairly rapid.

Three-dimensional zoom mode consists of live real-time 3D imaging in which the sonographer defines the size of a 3D echocardiographic sample volume by adjusting sample volume depth position, width, and height with a trackball and dials. Once the sample volume is defined, the image is flipped 90° in the vertical plane to demonstrate an en face view. Additional rotation of the image may be performed to achieve an anatomic orientation and view the structure of interest in real time. The volume rate is slow, typically from 5 to 18 volumes/sec.

Three-dimensional full-volume mode consists of four to seven live 3D contiguous sample volumes that are stitched together sequentially beat by beat over four to seven heartbeats into a single wide-angled, pyramid-shaped 3D sample volume. It is rapid reconstruction, not actual real-time viewing. It must be stored and cropped according to the particular region of interest. After the initial learning curve, this process takes approximately 1 min to perform most of the basic cropping functions. The volume rate is rapid, but the beat-by-beat electrocardiographically triggered reconstruction is prone to stitch artifact from either rapid heart rate or respiratory motion.

All three modes are selected with typical echocardiographic machine touch pad, dial, and trackball controls. All three modes can be further cropped in three orthogonal planes or using a single adjustable plane. Multiple adjustable parameters include gain, contrast, smoothing, image algorithms, color adjustments, color depth shading adjustments, rotation of images throughout 180° in orthogonal planes, and others. Image manipulation, cropping, and quantification can be performed on the ultrasound platform or on the digital storage and review station.

Multiple examples of images from each modality are provided in this report. In still images that are carefully acquired and cropped, it is not always apparent which 3D echocardiographic mode was used. In video images, the 3D zoom mode is noticeable by its slow volume rate and smooth images, and the 3D full-volume mode is noticeable by its stitch artifact if present.

Qualitative anatomic parameters delineated in each case included the type of ASD (secundum, primum, sinus venosus, common atrium, or coronary sinus), location, shape, and orientation. ASD shape was defined as oval, round, or triangular. Defects that were shaped somewhat like an egg or a pear, or slightly irregular, were considered oval shaped. ASD orientation was defined according to the long-axis orientation of the defect as vertical, horizontal, oblique with anterior tilt, or oblique with posterior tilt. Defects in which the lengths of the long-axis and short-axis dimensions were within 1 mm were designated as round.

Quantitative analysis of ASD by 3D echocardiography included maximum length, width, and area measured at atrial end-diastole. ASD dimensions were also measured at atrial end-systole to determine the change in dimensions over the cardiac cycle. ASDs were measured in an en face view from either the right atrial or left atrial perspective using dedicated quantitative software. Parameters calculated included the percentage change in ASD2 length, width, and area from atrial end-diastole to atrial end-systole. Atrial end-diastole was defined as the frame with the largest ASD2 dimensions and atrial end-systole as the frame with the smallest ASD2 dimensions. The number of holes in the atrial septum was quantified in all types. An ASD2 rim length of <5 mm was considered deficient. ASD2 rims were defined as follows: aortic rim = superior/anterior rim between the ASD2 and the aortic valve annulus; atrioventricular rim = inferior/anterior rim between the ASD2 and the atrioventricular valves; SVC rim = superior/posterior rim between the ASD2 and the SVC; inferior vena cava rim = inferior/posterior rim between the ASD2 and the inferior vena cava; and right pulmonary vein (RPV) rim = posterior rim between the ASD2 and the RPVs. In patients with ASD2s who underwent device closure, we compared 3D TEE–derived ASD length, width, and area at both end-systole and end-diastole with occlusive balloon size.

Statistical Analysis

Continuous data are reported as mean ± SD or as percentages as appropriate. The dimensions of ASDs and the change in dimensions during the cardiac cycle were analyzed by descriptive statistics, including minimum and maximum values, mean, standard deviation, median, and range. One-way analysis of variance with post hoc Tukey’s test was used to determine significant differences of continuous data between groups. Linear regression analysis was performed to determine the relationship between ASD balloon sizing and 3D echocardiographic parameters. Generalized linear modeling with multivariate analysis was used to determine the best or strongest 3D echocardiographic parameters to predict ASD balloon sizing. Bland-Altman analysis was used to examine intraobserver and interobserver variability. Statistical significance was declared when computed P values from two-sided tests were <.05. Analyses were performed using SPSS version 18.0 for Windows (SPSS, Inc., Chicago, IL).


This study was in compliance with all institutional guidelines regarding research conduct including patient confidentiality and institutional research board approval. All patients or parents had signed informed consent for complete 2D and 3D transesophageal echocardiographic studies.


Clinical Features and Feasibility

Patients ranged in age from 5 to 64 years (mean, 29.1 ± 18.2 years; median, 26.5 years), in body weight from 20 to 114 kg (mean, 64.2 ± 24.0 kg; median, 61.8 kg), and in body surface area from 0.8 to 2.4 m 2 (mean, 1.64 ± 0.38 m 2 ; median, 1.67 m 2 ). There were 45 female and 20 male patients. Device closure was performed in 44 patients, and 21 underwent surgery. No complications related to 3D TEE occurred.

Quantitative analysis was successfully performed in all 50 patients with ASD2s. The 3D echocardiographic modality that provided the best image for 3D quantitative analysis was 3D zoom mode in 36 patients (72%), full-volume 3D mode in seven (14%), and live 3D mode in seven (14%). The 3D zoom mode provided at least one image of sufficient quality for quantitative analysis in 48 patients (96%). The full-volume 3D mode provided at least one image of sufficient quality for quantitative analysis in 39 patients (78%), and the live 3D mode provided at least one image of sufficient quality for quantitative analysis in 31 (62%). The reasons for imaging failures were fast heart rate preventing consistent detection of atrial end diastole in two patients by 3D zoom, stitch artifact preventing clear border detection in 11 full-volume cases, and failure to define the entire defect borders in 19 patients by live 3D mode. Representative 3D echocardiograms of ASD2s are presented in Figures 1 A to 1 D (see Videos 1 and 2 , corresponding to Figures 1 C and 1 D [view video clip online]). Additional 3D echocardiograms of ASD2s with corresponding cardiac pathologic specimen examples are presented in Figure 2 (see Video 3 , corresponding to Figure 2 A [view video clip online]). Among the 15 patients with nonsecundum ASDs, quantitative analysis was successfully performed in 14 and was not applicable in the one patient with a common atrium.

Figure 1

Representative views and anatomic landmarks in an ASD2. (A) Right atrial (RA) and left atrial (LA) en face views. (B) Another example of RA and LA en face views. (C) Transgastric sagittal bicaval view acquired in live 3D mode from the standard perspective ( left ) (see Video 1 ) and posterior perspective ( right ). (D) Posterior-aspect views demonstrating the variable alignment between the septum primum (S1) and septum secundum (S2) over the cardiac cycle. ( Left ) Alignment between the septum secundum and S1 ( arrow ) components. Mild malalignment ( middle ) and more malalignment ( right ) between the septal components. As the malalignment increases, the size of the interatrial communication ( asterisk ) increases in size (see Video 2 ). In the orientation icon, blue designates the y plane, red designates the x plane, and green designates the z plane. A , Anterior; Ao , aorta; C , catheter; CS , coronary sinus; IVC , inferior vena cava; L , left; LPV , left pulmonary vein; P , posterior; R , right; RAA , right atrial appendage; RPA , right pulmonary artery; S , superior; TV , tricuspid valve.

Figure 2

Landmarks demonstrated from the RA perspective with pathologic specimen correlation. (A) Small ASD2 with atrial septal aneurysm acquired in 3D zoom mode (see Video 3 ). (B) Medium-sized ASD2 acquired in 3D zoom mode. (C) Large ASD2 acquired in 3D full-volume mode. In the orientation icon, blue designates the y plane, red designates the x plane, and green designates the z plane. AL , Anterior tricuspid valve leaflet; EV , Eustachian valve; RV , right ventricle; SL , septal leaflet of tricuspid valve. Other abbreviations as in Figure 1 .

Qualitative Anatomy

ASD Type

Fifty of 65 patients had ASD2s. Fifteen patients had ASDs of nonsecundum types, including seven with sinus venosus ASDs, six with primum ASDs (see Figure 3 A and Video 4 [view video clip online]), one with a common atrium, and one with a coronary sinus ASD. All 65 patients had correct ASD type diagnoses made with 3D TEE compared with 2D TEE as the reference parameter. Among the 21 patients who underwent surgery, 3D transesophageal echocardiographic and surgical inspection diagnosis of ASD type were concordant in all.

Figure 3

ASDs of nonsecundum type. (A) Primum ASD (ASD1), RA en face, four-chamber, and LA en face views (see Video 4 ). The ASD1 is located at the apical inferior margin of the atrial septum, with the atrioventricular (AV) valve as its lower margin. The long axis of the defect is oriented in an anterior-superior–to–posterior-inferior plane. Note in the four-chamber view that the AV valves are coplanar. (B) Superior sinus venosus ASD (SVASD) short-axis and long-axis views. The superior SVASD is located at the junction of the SVC to the atrium; the SVC overrides the defect. (C) Longitudinal view of a typical superior SVASD with anomalous connections of the RPVs to the SVC (see Video 5 ). Note that the SVC overrides the ASD (asterisk). (D) Longitudinal view of a typical inferior SVASD. Note that the IVC overrides the ASD ( asterisk ) (see Video 6 ). (E) Inferior SVASD en face views from the RA and LA perspectives. The defect is located at the cavoatrial junction. There is also an ASD2 present. Anomalous connections of RPVs to the IVC are often present, but not in this case. (F) Demonstration of a common atrium (CA) in a complex heart defect with right dominant AV septal defect, double-outlet RV, and RA isomerism. There is a band of tissue that traverses the CA, and no other atrial septal tissue is apparent. This collection of anatomic features is common in this complex anomaly. The AV valve is directed into the RV. ( Top middle ) Three-dimensional full-volume view with frontal cropping showing the CA, which is traversed by a thin band of tissue ( arrow ). ( Bottom left ) Three-dimensional zoom mode right-sided en face view of the band across the CA. ( Bottom right ) Corresponding left-sided en face view (see Video 7 ). (G) Coronary sinus ASD ( asterisk ) four-chamber view from a left anterior oblique perspective. The defect is oval in shape and located near the AV junction. The RA orifice is the enlarged coronary sinus orifice located in its usual position. The LA orifice is due to complete or partial unroofing of the coronary sinus into the left atrium. The AV valve relative positions and anatomy are normal, in contrast to ASD1 (see Video 8 ). In the orientation icon, blue designates the y plane, red designates the x plane, and green designates the z plane. LV , Left ventricle. Other abbreviations as in Figures 1 and 2 .

ASD Location

All 65 patients had correct ASD location diagnoses made with 3D TEE compared with 2D TEE as the reference parameter. Among the 21 patients who underwent surgery, 3D transesophageal echocardiographic and surgical inspection diagnosis of ASD location were concordant in all. Among the 50 patients with ASD2s, 31 defects were located in the central portion of the atrial septum, with all rims well developed. Among the other 19 ASD2s, 13 involved the central portion of the atrial septum but extended toward one of the rims, and six were positioned primarily toward the edge of the atrial septum. Among the 13 ASD2s with extension beyond the main central location, nine had extension toward the aortic valve (deficient aortic rim), three had extension toward the inferior vena cava (deficient inferior rim), and one had extension toward the SVC (deficient superior rim). Among the six ASD2s with off-center locations, three were positioned predominantly superior/anterior (deficient aortic rim), one was located superior/posterior (deficient SVC rim), one posterior (deficient RPV rim), and one inferior/posterior (deficient inferior vena cava rim).

Among the 15 patients with nonsecundum ASDs, seven had sinus venosus ASDs. Four were superior-type sinus venosus ASDs (see Figures 3 B and 3 C and Video 5 , corresponding to Figure 3 C [view video clip online]). Three were inferior-type sinus venosus ASDs (see Figures 3 D and 3 E and Video 6 , corresponding to Figure 3 D). Six patients had primum ASDs. One patient had a common atrium (see Figure 3 F and Video 7 [view video clip online]). One patient had a coronary sinus ASD (see Figure 3 G and Video 8 [view video clip online]).

ASD Shape

Among the 50 patients with ASD2s, oval shapes were present in 44, including three that were egg shaped and one that was pear shaped; the latter four were considered oval shaped for statistical analysis. Among the six remaining ASD2s, five were round and one was triangular. The seven sinus venosus defects, six primum ASDs, and the coronary sinus ASD were oval in shape. In the common atrium, the junction of the two sides of the atrium was round, except for the thin strand of traversing tissue. Among the 21 patients who underwent surgery, 3D transesophageal echocardiographic and surgical inspection results regarding shape were concordant in all. The shapes of the ASDs could not be reliably determined by 2D TEE.

ASD Orientation

In patients with ASD2s, oblique orientations of the ASD2 long axes were found in 20: 17 had the superior margin of the long axis tilted anterior, and three had the superior margin directed posterior. The orientation was vertical in 13 and horizontal in 12. Five defects were round and therefore lacked a long-axis orientation. In the nonsecundum ASD group, six of the seven sinus venous ASDs had horizontal long axes. In one patient with an inferior sinus venosus ASD, the defect was quite large, confluent with an ASD2, and had an oblique anterior-superior long-axis orientation. Primum ASDs had the long axis oriented in an anterior-superior–to–posterior-inferior direction. The common atrium had a round midportion and therefore no specific orientation. The coronary sinus ASD had a long-axis orientation in an anterior-superior–to–posterior-inferior direction. Among the 21 patients who underwent surgery, 3D transesophageal echocardiographic and surgical inspection results regarding shape were concordant in all.

Quantitative Anatomy

Quantitative ASD2 Anatomy

The largest ASD2 had a long-axis length of 4.6 cm, a short-axis length of 4.3 cm, and an atrial end-diastolic area of 14.2 cm 2 . The smallest defect measured 0.6 cm in the long-axis dimension and 0.3 cm in the short-axis dimension and had an area of 0.13 cm 2 . It was accompanied by an adjacent slightly larger defect (0.6 × 0.4 cm; area, 0.15 cm 2 ), both of which were closed by a single device. Details of the dimensions and dynamic change in dimensions of the three main types of ASD are presented in Table 1 .

Table 1

Dimensions and dynamic changes in dimensions for the three main types of ASDs

Atrial diastole Atrial systole Change in long axis Change in short axis Change in area
Variable Long (cm) Short (cm) Area (cm 2 ) Long (cm) Short (cm) Area (cm 2 ) cm % cm % cm2 %
ASD2 ( n = 50)
Minimum 1.00 0.30 0.24 0.60 0.10 0.09 0 0 0 0 0.10 9.10
Maximum 4.60 4.30 14.20 4.20 3.10 9.61 2.00 53.80 1.30 66.70 6.80 74.50
Range 3.60 4.00 13.96 3.60 3.00 9.52 2.20 73.80 1.40 86.70 6.70 65.40
Median 2.00 1.30 2.09 1.45 1.05 1.45 0.40 20.40 0.30 24.50 0.75 40.20
Mean 2.13 1.55 3.18 1.70 1.17 2.00 0.44 21.52 0.39 25.23 1.18 39.18
SD 0.94 0.84 3.12 0.87 0.66 2.03 0.35 14.68 0.32 16.22 1.31 16.35
Sinus Venosus ASD ( n = 7)
Minimum 2.30 1.60 2.96 2.30 1.40 2.83 0 0 0.10 4.30 0.13 4.40
Maximum 3.80 2.20 6.31 3.20 2.30 5.56 0.60 15.80 0.30 15.00 1.25 25.50
Range 1.50 0.60 3.35 0.90 0.90 2.73 0.60 15.80 0.20 10.70 1.12 21.10
Median 3.00 2.00 4.91 2.90 1.70 3.66 0.10 3.30 0.20 12.50 0.75 11.90
Mean 3.03 1.93 4.73 2.80 1.80 4.02 0.23 6.37 0.20 10.60 0.71 13.93
SD 0.75 0.31 1.68 0.46 0.46 1.40 0.32 8.33 0.10 5.60 0.56 10.70
Primum ASD ( n = 6)
Minimum 2.90 1.50 4.00 2.70 1.30 3.25 0.20 13.30 0.20 28.60 0.93 44.30
Maximum 1.50 0.70 0.70 1.30 0.50 0.39 0.10 4.50 0.20 13.30 0.31 18.75
Range 1.40 0.80 3.30 1.40 0.80 2.86 0.10 8.80 0 15.30 0.62 25.55
Median 2.20 1.40 2.59 2.10 1.20 1.66 0.20 6.90 0.20 14.30 0.75 35.90
Mean 2.20 1.20 2.43 2.03 1.00 1.77 0.17 8.23 0.20 18.73 0.66 32.33
SD 0.70 0.44 1.66 0.70 0.44 1.43 0.06 4.55 0 8.56 0.32 13.02

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Jun 15, 2018 | Posted by in CARDIOLOGY | Comments Off on Three-Dimensional Transesophageal Echocardiography of Atrial Septal Defect: A Qualitative and Quantitative Anatomic Study

Full access? Get Clinical Tree

Get Clinical Tree app for offline access