Sparse representation of deformable 3D organs

Parametric representation of deformable object with complex surface has been a challenge in various medical applications for its demanding resource consumptions. This paper proposed an efficient algorithm to construct a compact basis for a sequence of deformed 3D organ, in which those surfaces can be sparsely represented with a small number of parameters. The key idea in this paper is to explore the correlations among the deformed surfaces of an organ and extract the principle basis for representation and reconstruction. Both theoretical analysis and extensive simulations verified that the presented algorithm yields a three-order magnitude reduction in computational and storage complexity relative to traditional approaches while maintaining high precision for surface reconstruction. The proposed algorithm can be used for organ deformation tracking and optimal sampling strategy design.