Meeting Banner
Abstract #2619

Validation of Nonrigid Registration for Cardiac Cine MR Time Series

Guyu Li1, Li Zhang1, Christoph Guetter1, Christophe Chefd'hotel1

1Siemens Corporate Research, Princeton, NJ, United States


Validation of cardiac image registration has been a challenging task due to the lack of ground truth and severe through-plane motion. Since a consistent and transitive registration algorithm is more likely to find a biologically meaningful mapping, we focused on evaluating the consistency, transitivity, and consistency sensitivity to cardiac motion as the indicators of registration performance. A diffeomorphic nonrigid registration was compared with its symmetric version using 10 cardiac cine MR time series (SSFP, 1.5T) that were acquired from 10 volunteers. With the proposed criteria, we can give a more comprehensive picture of the registration performance on cardiac data sets.