Session S21.6

Semi-Automatic Detection and Tracking of Mitral and Aortic Annuli from Real–Time 3D Transesophageal Echocardiographic Images

F Veronesi*, C Corsi, V Mor-Avi, L Sugeng, EG Caiani,
L Weinert, C Lamberti, RM Lang

Università di Bologna
Bologna, Italy

The understanding of the morphology and function of normal mitral and aortic valves and their coupling is clinically important. To allow noninvasive evaluation of the two valves, a 3D imaging technique that provides simultaneous visualization of both valves is essential. The recently developed echocardiographic matrix array transesophageal (mTEE) transducer provides real-time 3D images of high spatial and temporal resolution that may be suitable for detailed study of functional anatomy of the two valves. To test this hypothesis, we developed and tested a tool to detect and track throughout the cardiac cycle mitral and aortic annuli (MA and AoA) on mTEE images. Methods. We used mTEE images acquired during clinically indicated studies in 15 patients with normal valves. To detect the two annuli, 15 planes (24° apart) rotated around each valve’s axis were displayed. For every plane, two points were initialized on the valve annulus. Polynomial interpolation was used to show in 3D space the splines defining MA and AoA. To track the position of all initialized points, we implemented a two-step 3D feature tracking algorithm: first, the likelihood between the neighbours of each point in the initial and the following frames was maximized, then Lucas-Kanade optical flow algorithm was applied to improve the accuracy of tracking. Finally, the points tracked throughout the cardiac cycle were interpolated and displayed. Manual corrections were performed when necessary. MA surface area, AoA projected area and MA-AoA angle were automatically measured from the detected annuli throughout the cardiac cycle. Results. Our software allowed frame-by-frame detection and tracking of MA and AoA in all study patients. Maximum and minimum values were 9.2±2.0 cm2 and 7.6±2.0 cm2 for MA surface area, and 4.9±1.3 cm2 and 3.8 ± 1.0cm2 for AoA projected area, respectively. When MA surface area was maximal during early diastole), AoA projected area was at its minimum, and vice versa during systole. The minimum value for the angle between MA and AoA occurred at end-systole and was significantly smaller than the angle at end-diastole. In summary, our software allowed for the first time non-invasive quantitative measurements of the 3D dynamic geometry of normal MA and AoA and their coupling from mTEE data.

(Abstract Control Number: 287)