Canadian Cardiovascular Society
Montreal, October 29-November 2, 1996

3D ECHOCARDIOGRAPHY BY RAPID FREE SCANNING FROM MULTIPLE TRANSTHORACIC WINDOWS

D. Leotta, B. Munt, M. Legget, R. Martin, X.-N. Li, E. Bolson, T. Zwink, C. Kraft, G. Bashein, M. Sivarajan, S. Vaezy, C. Otto, F. Sheehan, Seattle, WA


Three-dimensional (3D) echocardiography has recently been shown to improve the accuracy and reproducibility of estimates of left ventricular (LV) mass and volume. However, accuracy of 3D cardiac reconstructions can be limited by such factors as lengthy image acquisition times, limited acoustic viewing windows, a need for respiratory gating, and geometric assumptions of LV shape.

We present a method for acquiring a 3D scan in less than 1 minute using a commercial scanner equipped with a magnetic position locating device (Ascension Technology Flock of Birds). We image continuously while freely tilting, translating and/or rotating the probe in held respiration. In three breath holds (8-13 sec each), complete scans from the parasternal (short and long axis) and apical windows are obtained.

Images are digitized into files linked to their spatial coordinates and the electrocardiogram. Object borders are manually traced using custom software that allows interactive 3D visualization and editing of the outlines. The scanned objects are reconstructed using a piecewise-smooth subdivision surface-fitting algorithm.

Phantom imaging indicates that the system accurately reproduces volume (true volume = 0.97 calculated volume + 1.4 ml, r2=0.999, SEE=1.7 ml, n=10 balloons) and shape (rms distance of reconstructed surface to ideal surface < 1 mm, n=3 cylinders). Initial scans in humans indicate LV reconstructions are feasible in awake patients.

These results suggest that LV volume and shape can be calculated accurately from images acquired during rapid free scanning in vivo. Reconstruction of the end diastolic and end systolic LV can be performed quickly from one or more acoustic windows without respiratory gating or geometric assumptions. We advocate further evaluation of this technique for 3D scanning.