Monogenic phase based myocardial motion analysis from cardiac ultrasound with transverse oscillations

We present a motion estimation algorithm for cardiac ultrasound images with transverse oscillations after baseband demodulation in the axial direction (B-mode TO). Our algorithm computes the motion by solving the optical-flow equation locally on a sliding spatial window. Conservation of the monogenic phase is assumed in lieu of the traditional pixel brightness. A local affine model accounts for typical cardiac motions as contraction/expansion and shearing. A coarse-to-fine B-spline scheme allows a robust and effective computation of the model parameters. To improve the precision of the estimate, we adopt a pyramidal refinement scheme based on the multiband nature of B-mode TO images: the lowest harmonic is used to produce a rough estimate of the displacement; higher harmonics are then used to refine it thanks to their higher level of detail. Performance was evaluated on realistic simulated cardiac ultrasound sequences and compared to a well-established algorithm. With respect to the latter, our algorithm produced an average reduction of 40.8% of the endpoint error. Also, the contribution of the higher lateral harmonics typical of B-mode TO images was shown to considerably improve the estimation accuracy.