Elastic & Efficient Three-Dimensional Registration for Abdominal Images

Biomedical image registration plays an important role in the intelligent use of large amount of medical data to establish safe healthcare. One of main challenges in abdominal image registration is to deal with elastic deformations caused by the shape and volume changes of internal organs, as well as heartbeat and breath. Other challenges include high computational efficiency required for clinical applications and accurate registration for 3D multimodality images. In this paper, an elastic, efficient and automatic 3D abdominal image registration method is proposed. Our proposed algorithm is divided into two stages and applied to register slice pairs of 3D volume data. First, non-iterative affine registration is performed to fix the global deformation by minimizing mean squared error (MSE). Then, effective active contour-based elastic registration is performed to correct the local deformations. This method has been validated by the experiments on both 3D CT-CT registration and CT-PET registration