Accurate diagnosis of mitral regurgitation (MR) severity is central to proper treatment. Although numerous approaches exist, an accurate, gold-standard clinical technique remains elusive. The authors previously reported on the initial development and demonstration of the automated three-dimensional (3D) field optimization method (FOM) algorithm, which exploits 3D color Doppler ultrasound imaging and builds on existing MR quantification techniques. The aim of the present study was to extensively validate 3D FOM in terms of accuracy, ease of use, and repeatability.
Three-dimensional FOM was applied to five explanted ovine mitral valves in a left heart simulator, which were systematically perturbed to yield a total of 29 unique regurgitant geometries. Three-dimensional FOM was compared with a gold-standard flow probe, as well as the most clinically prevalent MR volume quantification technique, the two-dimensional (2D) proximal isovelocity surface area (PISA) method.
Overall, 3D FOM overestimated and 2D PISA underestimated MR volume, but 3D FOM error had smaller magnitude (5.2 ± 9.9 mL) than 2D PISA error (−6.9 ± 7.7 mL). Two-dimensional PISA remained superior in diagnosis for round orifices and especially mild MR, as predicted by ultrasound physics theory. For slit-type orifices and severe MR, 3D FOM showed significant improvement over 2D PISA. Three-dimensional FOM processing was technically simpler and significantly faster than 2D PISA and required fewer ultrasound acquisitions. Three-dimensional FOM did not show significant interuser variability, whereas 2D PISA did.
Three-dimensional FOM may provide increased clinical value compared with 2D PISA because of increased accuracy in the case of complex or severe regurgitant orifices as well as its greater repeatability and simpler work flow.
The course of treatment for mitral regurgitation (MR) is determined according to valve morphology and MR severity. Surgery is indicated only for severe MR, yet a significant number of patients referred to surgical centers for severe MR on the basis of echocardiography are found to have only mild or moderate MR on subsequent evaluation. This reality places critical importance on diagnostic accuracy and reproducibility. The persistent lack of a “gold-standard” diagnostic tool compounds the clinical burden of MR.
Among many quantitative criteria that have been proposed to aid diagnosis, the two-dimensional (2D) proximal isovelocity surface area method (PISA) has become the most common tool in the past two decades. Two-dimensional PISA assesses MR by quantifying effective regurgitant orifice area (EROA) and regurgitant volume (RVol). However, much like other metrics, 2D PISA’s technical limitations diminish its accuracy and repeatability. Its assumption of a true hemispheric convergence zone loses validity for slit-like orifices, which are more common in severe disease. Additionally, the measured velocity is a projection of the true maximum velocity, which underestimates the maximum velocity according to the cosine of the angle between the ultrasound beam and maximum velocity vector.
Three-dimensional (3D) ultrasound offers potential for new MR quantification techniques. Although a few such techniques have been proposed, some concerns exist surrounding their accuracy and/or viability. A proposed 3D PISA tool, for example, shows initial promise, but the need for time-consuming offline manual correction of automated measurements may pose a challenge. Our group recently developed a novel approach to MR quantification by 3D ultrasound. The 3D field optimization method (FOM) quantifies RVol in a manner that addresses key limitations of alternative techniques. A curve-fitting process estimates the maximum regurgitant velocity. Three-dimensional FOM’s automated processing algorithm does not require manual precision in measuring the PISA region and requires minimal offline processing. We previously described the development of the 3D FOM algorithm, including its initial validation using idealized, static round and slit orifices machined into a rigid plate. Three-dimensional FOM was shown to perform well with both orifice types and to outperform 2D PISA when applied to a slit orifice.
The meaningful impact an improved MR quantification technique would have on MR management is evident. On account of the novel features of 3D FOM, we predict that it may provide superior utility to alternative techniques. Before 3D FOM can be adopted clinically, it must be tested on regurgitant mitral valves (MVs) and evaluated against a gold-standard RVol measurement. Therefore, in the present study we applied 3D FOM to explanted ovine MVs, mounted in an ex vivo left heart simulator, under a range of tightly controlled MR flows. The setup was equipped with a gold-standard flow probe. We concretely evaluated 3D FOM against both 2D PISA and the flow probe, in terms of ease of use, accuracy, and repeatability.
Ovine hearts were acquired from a local market. MVs were explanted and mounted in the extensively studied Georgia Tech Left Heart Simulator, as previously described ( Figure 1 A). Using this closed-loop system, healthy human left heart hemodynamics were established for each valve (cardiac output 5.0 L/min, heart rate 70 beats/min with 70% diastole, peak transmitral pressure 120 mm Hg). Zero or trace MR was observed at this stage. Subsequently, controlled states of annular dilatation, papillary muscle displacement, and systemic hypertension (increased resistance) were applied to yield a spread of regurgitant orifices ( n = 29 simulated regurgitant MVs, from five original ovine valves). Mitral flow rate was recorded at 1 kHz using an analog electromagnetic flow probe (200 series; Carolina Medical Equipment, East Bend, NC) upstream of the left atrium. En face high-speed imaging of the valve (A504K; Basler AG, Ahrensburg, Germany) was performed.
All echocardiographic imaging was performed using an iE33 with an X7-2 matrix-array probe (Philips Health Tech, Andover, MA). The probe was mounted on a traverse, positioned en face to the valve, and aligned with the largest 2D PISA jet for each sample. Three-dimensional color images were acquired over seven heartbeats, with frame rates of 20 to 22 Hz. Grouping of orifice shape as “round” or “slit” used high-speed images of the valve in combination with the ultrasound C-plane 1 to 3 mm subannularly ( Figure 2 ). Slit orifices had aspect ratios of 2:1 or greater at peak systole. Per American Heart Association guidelines, samples were diagnosed as having mild (RVol < 30 mL), moderate (RVol ≥ 30 mL but <60 mL), or severe (RVol ≥ 60 ml) MR.
The flow probe used in this study was calibrated over a range of flow rates ( R 2 > 0.98, accuracy ±3.5%). To derive a true MR value for each valve, the flow profile from the probe was analyzed. The retrograde flow phase of the cardiac cycle was partitioned into closing and leakage phases, as depicted in Figure 1 B. True RVol was defined as the integral of the mitral flow rate over the leakage phase only.
Two-dimensional PISA and 3D FOM were performed as previous published. In brief, RVol was calculated by 2D PISA using the EROA as follows:
EROA = 2 π r 2 × V alias V max
MR volume = EROA × velocity-time integral