Background
The pulmonary valve (PV) is rarely visualized in short axis with conventional two-dimensional transthoracic echocardiography (TTE). Thus, the true incidence of abnormal PV morphology in patients undergoing TTE is unknown. This study sought to evaluate the feasibility of using three-dimensional echocardiography in the morphologic assessment of the PV in short-axis.
Methods
A total of 200 consecutive patients referred for routine TTE were prospectively evaluated (mean age 64 ± 16 years; 113 males). Live3D and full-volume 3D (FV3D) were performed with the feasibility of visualizing PV morphology assessed. McNemar’s test was used as a nonparametric comparator between Live3D and FV3D results and to assess for any significant learning curve. Chi-square test was used to determine the association between variables.
Results
PV morphology detection rates were significantly different ( P < . 0001) between Live3D (60%) and FV3D (23%). The optimal plane for Live3D was the parasternal view (99%), using zoom over the PV and rotating to a short-axis image. PV short-axis cusp detection using Live3D was dependent on the initial two-dimensional PV image quality ( P < . 0001).
Conclusion
Live3D is feasible in evaluating PV short-axis morphology and provides incremental value in the TTE examination.
Assessment of pulmonary valve (PV) morphology is challenging using conventional two-dimensional transthoracic echocardiography (TTE) as short-axis visualization rarely allows the valve cusps to be seen in their entirety. Typically, only two of the cusps are able to be visualized simultaneously ( Figure 1 ), and therefore the true incidence of abnormal PV morphology diagnosed by TTE is unknown.
Three-dimensional TTE is emerging as a clinically useful tool providing incremental information and complementing the routine TTE examination. Three-dimensional TTE provides a more accurate visualization of anatomic structures because it presents a complete view of the heart from multiple perspectives, enabling visualization of intracardiac structures from virtually any view. Three-dimensional TTE may enhance the understanding of PV morphology by enabling short-axis views of this valve that are not obtainable using two-dimensional TTE.
This study sought to determine the feasibility of using three-dimensional TTE to assess PV morphology in the short-axis view. In addition, the study sought to determine whether the rate of visualization was different between the two different three-dimensional imaging modalities: Live3D and full-volume 3D (FV3D) imaging. Finally, the presence of any significant learning curve was also determined.
Materials and Methods
A total of 200 consecutive patients (mean age 64 ± 16 years; 113 males) who were referred for TTE as part of their routine clinical care were prospectively evaluated. Patient demographic data are presented in Table 1 . None of the patients had significant PV disease visualized from the conventional two-dimensional and color Doppler images. Imaging was performed with a Philips iE33 ultrasound system and x3-1 matrix array transducer (Philips Medical Systems, Andover, MA) using both Live3D (one cardiac cycle) and FV3D (with electrocardiogram triggering over four cardiac cycles to create a larger volume of data) acquisition modes. Because Live3D obtains images from one cardiac cycle, it offers artifact-free instantaneous imaging but is limited by a narrow angle with a partial volume of data from the heart. In comparison, FV3D gathers multiple volumes of data from multiple cardiac cycles to create a pyramid of data that are usually large enough to capture the entire heart. FV3D relies on stitching together these volumes acquired from different consecutive cardiac cycles and can therefore be limited by “stitch artifacts.”
| Parameter | |
|---|---|
| Age (y) | 64 ± 16 |
| Gender | |
| Male | 113 (56.5%) |
| Female | 87 (43.5%) |
| Height (cm) | 168 ± 10 |
| Weight (kg) | 79 ± 18 |
| BSA (m 2 ) | 1.91 ± 0.26 |
| Heart rate (bpm) | 71 ± 15 |
| Rhythm | |
| Sinus | 167 (83.5%) |
| Sinus with ectopy | 5 (2.5%) |
| Atrial fibrillation | 22 (11%) |
| Paced | 5 (2.5%) |
| Indeterminate | 1 (0.5%) |
Echocardiographic Study
All patients underwent a comprehensive TTE study lying in the left lateral decubitus position, incorporating two-dimensional, Doppler, and color Doppler imaging according to the American Society of Echocardiography recommendations. Three-dimensional imaging was performed at the end of the routine study, with images acquired from the parasternal (long- or short-axis) and apical four-chamber views. Three-dimensional imaging was not performed from the subcostal view because the resolution was deemed suboptimal. Live3D assessment incorporated Live3D zoom to allow a more focused visualization of the PV and elimination of myocardium and other intracardiac structures that impeded a short-axis visualization. FV3D datasets were acquired during a breath-hold and gated over four to seven cardiac cycles. Final volumes of data were stored on a digital archive and evaluated offline using QLAB 7.0 software (Philips Medical Systems Andover, MA) to obtain a short-axis view of the PV. Three-dimensional gain and brightness were adjusted with XRes Adaptive Image Processing (provides image enhancement to reduce speckle, haze, and clutter artefacts, and enhances edges) set to low to improve delineation of anatomic structures. Vision H (color map to enhance depth perception by coding nearer structures orange and structures further away blue) was used to improve visualization of valvular detail.
Three-dimensional studies were performed and analyzed by a single reviewer experienced in the acquisition and interpretation of three-dimensional echocardiographic images. PV morphology image quality in two-dimensional, Live3D, and FV3D was graded and classified using a semiquantitative four-point scale: poor (unable to visualize the PV cusps); fair (able to visualize number of leaflets and coaptation); good (good visualization of the number of leaflets and their coaptation and mobility); and excellent (excellent visualization of the number of leaflets and their coaptation, mobility, and thickness).
Statistical Analysis
Data were collected and statistically analyzed using MedCalc Version 10.4.5 statistical software (Mariakerke, Belgium). Numeric variables were expressed as mean ± 1 standard deviation. Means were compared using an unpaired t test. A P value of < .05 was considered significant. McNemar’s statistical test was used as a nonparametric comparator between the rate of PV morphology detection by Live3D and FV3D. McNemar’s statistical test was also used to detect a significant learning curve by comparing the results obtained between the first and second 50 patients, as well as the first and second 100 patients. To prove whether the initial two-dimensional image quality of patients influenced whether the PV was visualized by Live3D as well as the resultant three-dimensional image quality, a chi-square test was used.
Results
By using Live3D, the PV morphology ( Figure 2 ) was visualized sufficiently with the number, thickness, and mobility of the cusps seen in 118 patients (60%). The optimal plane for Live3D imaging was the parasternal view (99%) using Live3D zoom over the PV and rotating to a short-axis image. The image quality was excellent in 1 patient (0.5%), good in 14 patients (7%), fair in 103 patients (51.5%), and poor in 82 patients (41%) ( Figure 3 ).
By excluding patients with technically difficult initial two-dimensional imaging quality (n = 46) from the Live3D cohort, the detection rate of PV morphology increased to 76.6%. Of the patients in whom the PV cusps were visualized adequately by Live3D, all appeared trileaflet. Because Live3D is not limited by the stitch artefact (one volume of data), it was hypothesized that after excluding patients with technically difficult initial two-dimensional images, this imaging modality may be limited by operator experience. Notably, after the patients with technically difficult two-dimensional images were excluded, the detection rate in the first 100 patient group was 58.4% (32 patients), which significantly increased ( P < . 0001) to 94.8% (73 patients) in the second 100 patient group.
In comparison, the rate of visualization was remarkably lower with the FV3D acquisition mode detecting adequate PV morphology in only 44 patients (22%). Consequently, there was a significant difference ( P < . 0001) in the rate of visualization of the PV in short axis between the two three-dimensional imaging modalities. The main limitation for FV3D imaging was related to poor resolution because of technically difficult initial two-dimensional images (23%) or dropout (49.5%) (patient’s body surface area, lung disease, and large footprint of the transducer). Second, it was related to stitch artefact (16.5%) caused by rhythm disturbances (atrial fibrillation, ventricular ectopy) or patients unable to breath-hold (11%). There was no significant change in the detection rate between the first and second 100 patients in the FV3D group.
The ability to visualize the PV in its short axis by Live3D was related to the initial two-dimensional PV image quality ( P < . 0001, contingency coefficient 0.564). As shown in Figure 4 A, if the initial quality of the PV on the two-dimensional images was technically difficult, the chances of detecting the PV in short-axis by Live3D was limited. Once the initial two-dimensional image quality of the PV became fair, the chances of visualizing the PV increased.
