ML-based Motion Estimation in Ultrasound Images Using Heavy-tailed Noise Distributions

Nora Ouzir, Esa Ollila, Sergiy A. Vorobyov

Abstract: The multiplicative Rayleigh noise model has been used for maximum likelihood (ML) motion estimation in ultrasound imaging (UI). In this work, we introduce new robust similarity measures that take into account the deviations from the Rayleigh statistics resulting, for example, from multiple scatterings or acquisition artefacts. Specifically, the t-distribution is used for modelling the radio-frequency (RF) signals and the Nakagami-Gamma (NG) model is used for the echo amplitudes. Experiments using in vivo images of the carotid artery show an improvement in motion estimation accuracy in comparison with the similarity measure based on the classical Rayleigh model.

External IDs:dblp:conf/camsap/OuzirOV19