Improved accuracy of noninvasive flow rate calculation by three-dimensional reconstruction of digital velocity maps: In vitro and in vivo validation in patients with mitral stenosis

Recent proximal acceleration technique for calculating flow rate through restrictive orifices assumes a hemispherical isovelocity contour shape for extrapolation from 2D images to 3D surface area. This restricts clinical implementation because isovelocity contour shape varies. We therefore tested the hypothesis that 3D reconstruction of digital velocities more accurately measures flow through restrictive orifices than standard 2D techniques because it provides surface area directly without assumptions or inlet angle correction. This was first tested in mitral stenotic (MS) models produced from 3D patient studies by stereolithography (3 shapes, .5-1.5 cm² orifices, 28 flow rates), and then in 25 consecutive patients with MS undergoing catheterization (≤ trace regurgitation), integrating forward flow throughout diastole. In vitro, flow rate from reconstructing 6 rotated views correlated and agreed well with actual values (r=.99, SEE=5 ml/sec, mean error=-2±5ml/sec), with virtually no difference using only 2 views. Scatter was significantly lower than by 2D techniques, which systemically underand overestimated at varying flow rates, Nyquists, and distances from the orifice (SEE=26ml/sec, p<.01 vs 3D; mean error=-6±37ml/sec). In patients, results from reconstructing two perpendicular views also correlated and agreed well with invasive values for cardiac output (r=.96 SEE=0.31/min. mean error=0±0.281/min) and effective orifice area (r=.89, SEE=0.12cm², mean error=-0.05±0.13cm²). Conclusion: The accuracy of the proximal flow convergence method can be improved by 3D velocity reconstruction without geometric assumptions. This approach is clinically feasible with existing equipment using only 2 views, and is suitable for automatization to provide a powerful clinical tool.