Affiliations: [a] Department of Computer Science, University of New Orleans, New Orleans, LA, USA | [b] Department of Biomedical Engineering, University of Alabama at Birmingham, Birmingham, AL, USA | [c] Department of Medicine, Allegheny General Hospital, Pittsburgh, PA, USA
Correspondence:
[*]
Address for correspondence: Eduardo Kortright, Ph.D., Computer Science Department, Mathematics Building 312-D, University of New Orleans, New Orleans, LA 70148, USA. Tel.: +1 504 280 6626; Fax: +1 504 280 7228; E-mail: eduardo@cs.uno.edu.
Abstract: Control-volume (CV) methods applied to magnetic resonance velocity-encoded cine images of the convergent proximal flow field of a regurgitant valve have been shown to measure reverse blood flow volume accurately. Spatial and temporal averaging are known to affect accuracy, but the effects of slice thickness and orientation relative to the flow field have not been systematically studied, nor have CV configurations requiring fewer scans been explored. Further, surface area calculations at the intersection of CV walls are a previously unrecognized source of error. Using a computational fluid dynamics model of steady flow through an orifice, we evaluated five different CV configurations in terms of accuracy, time costs, and clinical potential. CVs incorporating a basal wall were affected by blurring of axial velocity gradients near the orifice, and voxel grid alignment relative to the orifice was the most significant source of inaccuracy. Errors in surface area calculations at plane intersections produced deviations of 7–20%, depending on configuration. A CV formed by slices parallel to the orifice plane was deemed clinically unusable, while a cylindrical CV yielded good accuracy in simulated tests and showed potential for practical implementation based on scan time, ease of view selection, and visualization of the flow field.
