Non-parametric Image Registration Using Generalized Elastic Nets

We introduce a novel approach for non-parametric non-rigid image registration using generalized elastic nets. The concept behind the algorithm is to adapt an elastic net in spatial-intensity space of one image to fit the second image. The resulting configuration of the net, when it achieves its minimum energy state, directly represents correspondence between images in a probabilistic sense and recovers underlying image deformation, which can be arbitrary. Representation of elastic net in the spatial-intensity space with specific priors that enforce natural elastic deformation is introduced. Efficient algorithm for optimization of elastic net energy is developed. The accuracy and effectiveness of the method is demonstrated on different medical image registration examples with locally non-linear underlying deformations.