Multiview diffeomorphic registration for motion and strain estimation from 3D ultrasound sequences

摘要: This paper presents a new registration framework for estimating myocardial motion and strain from multiple views of 3D ultrasound sequences. The originality our approach resides in the estimation transformation directly rather than single view or reconstructed compounded sequence. allows us to exploit all spatio-temporal information available input avoiding occlusions image fusion errors that could lead some inconsistencies quantification result. In addition, by using original images, speckle (which is an important feature be blurred out process) should remain consistent between temporal frames. We propose multiview diffeomorphic strategy enforces smoothness consistency domain modeling continuous 3D+t velocity field as sum B-spline kernels. representation robustly cope with variations heart rate resulting different number images acquired per cardiac cycle views. similarity measure obtained extension pairwise mean square error metric where weighting scheme balances contribution views. We have carried experiments on synthetic known ground truth in-vivo data sets two volunteers. It shown inclusion several improves curves reduces segments non-physiological pattern observed.