Mid Sweden University

Major cardiac interventions are known to have a negative impact on heart function. Patients that undergo cardiac surgery have their heart carefully monitored perioperatively in order to determine whether the heart restore and maintain desired functionality. Currently, cardiac monitoring remains largely qualitative. Quantitative measures for cardiac function has been an active field of research for years and with technological advancement being made to transesophageal echocardiography, it is expected that automated functional monitoring can be facilitated. TEE provides acquisitions of relatively high- quality and frame rate, making the computation of both global and regional cardiac function feasible. Mitral annular systolic excursion, mitral annular peak velocities and strain are well established parameters that describe the motion, rate of motion and deformation of the myocardium, and they are reported to relate well with cardiac function.

The proposed methods compute cardiac function parameters by a combination of speckle tracking and velocity estimation from tissue Doppler measurements. Speckle tracking is implemented with a block matching approach in order to estimate the magnitude and direction of motion of cardiac structures between frames. While speckle tracking determine movement, velocity estimation determine the rate of motion. Traditional tissue Doppler techniques are angle dependant, only measuring velocity in beam direction, however by combining the velocity measurements with the speckle tracking implementation, true velocity of the cardiac structures are estimated.

The implementation have been executed on in vivo acquisitions of B-Mode images and tissue Doppler measurements for the three cardiac planes; 4 chamber, 2 chamber and long axis, recorded from five patients that all underwent variations of bypass surgery. The results from the application have been compared with manually derived ground truth measurements. The results show that MAPSE was derived with a mean difference of 4,4%, for viable cardiac regions. For practical reasons it is not possible to align the TEE transducer with the left ventricle's apex to base axis, and therefore foreshortening will occur. This is most prominent from the estimated mitral annular peak velocities, where the 4 chamber plane show a 18,5% increase in S' and a 15,5% increase in E', whilst the results from the 2 chamber estimates show an increase of 188,5% in S' and a 55% increase in E'. The computation of Lagrange strain showed an mean increase of 6% to the reference measurements. The results show great promise when keeping in mind resolution, intra-observer variability and the different methods used for the measurements and the reference values. The application is somewhat limited by acquisition. Out-of-plane motion, unstable speckle patterns and loss of anatomical landmarks limits the full potential of a speckle tracking approach.