Spatiotemporal Atlas Estimation for Developmental Delay Detection in Longitudinal Datasets
MICCAI '09 Proceedings of the 12th International Conference on Medical Image Computing and Computer-Assisted Intervention: Part I
Population Shape Regression from Random Design Data
International Journal of Computer Vision
Schild's Ladder for the parallel transport of deformations in time series of images
IPMI'11 Proceedings of the 22nd international conference on Information processing in medical imaging
Hi-index | 0.00 |
The present paper builds upon recent advances in the spatiotemporal alignment of cardiac sequences to construct a statistical atlas of normal motion. Comparing cardiac sequences requires considering both the temporal component (changes along the sequences) and the inter-subject one. The objective here is to understand the changes in the comparison of myocardial velocities depending on (1) the chosen reorientation action (finite strain [local rotation only], local rotation and isotropic scaling, or full Jacobian matrix using the push-forward) and (2) the chosen system of coordinates (Lagrangian, Eulerian, or if a compromise between both [e.g. hybrid-Eulerian] is possible). Myocardial velocities are estimated locally using speckle tracking on echocardiographic (US) sequences, then aligned to a reference timescale, and finally reoriented to the anatomical reference according to the chosen reorientation framework. The methodology was applied to 2D US sequences in a 4-chamber view from 71 healthy volunteers. Experiments highlight the limitations of the hybrid-Eulerian scheme, showing that the intra-subject transformation should be taken into account, and discuss the options to perform the inter-subject one.