Comparing Two‐Dimensional Ellipsoid Model Variants in Estimating Three‐Dimensional Echocardiographic Right Ventricular Volume in Dogs

Determining right ventricular (RV) volume typically requires three-dimensional imaging due to its complex shape. The ellipsoid model offers a two-dimensional alternative, employing area- or linear-based formulas with further variations depending on the echocardiographic views used for measurements. To identify which ellipsoid model variant best agrees with real-time three-dimensional echocardiography (RT3D) as a reference standard and to assess within-day reproducibility. Sixty-seven client-owned dogs (23 normal, 44 with right-sided heart diseases) underwent echocardiograms, with 20 normal dogs receiving a repeat examination. Prospective method comparison study. Body weight-indexed end-diastolic (iEDV) and end-systolic volumes (iESV) were calculated across eight ellipsoid model variants. Agreement was assessed using concordance correlation coefficients (rc) and Bland-Altman analysis, while within-day reproducibility was evaluated using intraclass correlation coefficients (ICC) and reproducibility coefficients. The area- and linear-based variants using RV parameters from the left apical four-chamber view and perpendicular diameter from the right parasternal short-axis view (AEMA4C-RPS and LEMA4C-RPS, respectively) were the only methods to achieve moderate agreement with RT3D (rc > 0.90). The AEMA4C-RPS showed no significant systematic bias for iEDV (median of the differences [95% confidence interval]: 0.09 [0.00-0.13]), while LEMA4C-RPS exhibited no significant systematic bias for iEDV (0.03 [-0.02-0.08]) and iESV (0.04 [-0.01-0.12]), though biases increased at larger volumes. Both methods demonstrated good reproducibility (ICC > 0.75), with iESV reproducibility significantly greater than that of methods using RV parameters from the right parasternal long-axis four-chamber view. The AEMA4C-RPS and LEMA4C-RPS offer practical RV volume estimates.