A new variational technique for combining affine registration and optical flow in echocardiography images

Since myocardial motion is directly related to cardiac vascular supply, it can be helpful in diagnosing the heart abnormalities. The most comprehensive and available imaging study of the cardiac function is B-Mode echocardiography. However diagnostic systems are expert dependent and motion is not clear in the B-mode echocardiography images and therefore many efforts are pointed toward proposing new methods to measure the motion accurately. So far there have been many methods for myocardial motion estimation such as affine registration, elastic registration or optical flow but each method suffers from lack of accuracy. To increase the accuracy of motion detection techniques, we propose a new algorithm based on a variational technique to combine the efficiencies of optical flow methods and affine registration in combination with multi-resolution spatiotemporal Spline moments. The evaluation was performed on simulated, synthetic and real data. A comparison between the proposed method and several other methods shows its better performance to measure the motion more accurately in the presence of shear, rotation and noise. The proposed method achieved rotational error of 2.6 degrees per frame and amplitude error of 3.7 percent per frame. These results demonstrate a better efficiency with respect to other B-Mode echocardiography motion estimation techniques such as Lucas-Kanade, Horn-Schunck and spatiotemporal affine technique.