A comparison of two analytical systems for 3-D reconstruction from biplane videoradiograms

The authors compare the accuracy and reproducibility of two systems for measuring three-dimensional motions of the human heart from biplane videoradiograms of surgically inserted radiopaque markers. System 1 uses orthogonal imaging systems with coincident perpendicular central rays and a modified parallel-ray approximation. System 2 uses a mathematical determination of the geometry of the imaging systems each time a sequence is recorded. Accuracy of the calculated lengths, expressed as mean percentage error of a series of known lengths, is <0.7% for system 1 and 0.2% for system 2; reproducibility is 3.1% for system 1 and 1.0% for system 2. In both systems, reproducibility is limited by resolution of the imaging-recording-digitizing chain. The authors conclude that both systems could be improved by increasing the effective digitizer resolution, either by making the image larger or by using digitizing hardware with greater resolution, or both. Because system 2 provides improved accuracy and reproducibility and geometrical alignment requirements are minimal, it has been selected for biplane image data acquisition and three-dimensional reconstructions.<
>