In vivo tracking of 3D organs using spherical harmonics and subspace clustering

Deformable organ tracking has been a challenge in various medical applications. This paper proposes an algorithm for 3D organ tracking based on spherical harmonics (SH) and subspace clustering. The potential deformation subspaces are identified from training data, based on which an extremely low density sampling strategy and a low cost deformation construction method are designed. Both theoretical analysis and simulations verified that the presented tracking algorithm minimizes the number of sampling locations, storage and computation complexity, while maintaining high accuracy. The designed approach can be applied to in vivo 3D organ tracking and visualization during surgical intervention.