A symmetric deformation-based similarity measure for shape analysis

Statistical modeling of cellular/subcellular shapes is required for quantifying shape variations and understanding complex cell behaviors. Such statistical models rely on the choice of a proper similarity measure. In this paper we introduce a symmetric deformation-based similarity measure for cellular shape analysis. The proposed method is based on finding an optimal diffeomorphic mapping between shape images that minimizes a physically meaningful energy function. We compare our proposed method to the large deformation dif-feomorphic metric mapping (LDDMM) and the standard Euclidean metric on a nuclei dataset and show that the proposed method outperforms the others in capturing the variations in the dataset.