Dynamic Multi-object Gaussian Process Models

Abstract : Statistical shape models (SSMs) are state-of-the-art medical image analysis tools for extracting and explaining shape across a set of biological structures. A combined analysis of shape and pose variation would provide additional utility in medical image analysis tasks such as automated multi-organ segmentation and completion of partial data. However, a principled and robust way to combine shape and pose features has been illusive due to three main issues: 1) non-homogeneity of the data (data with linear and non-linear natural variation across features), 2) non-optimal representation of the 3D Euclidean motion (rigid transformation representations that are not proportional to the kinetic energy that moves an object from one position to the other), and 3) artificial discretization of the models. Here, we propose a new dynamic multi-object statistical modelling framework for the analysis of human joints in a continuous domain. Specifically, we propose to normalise shape and dynamic spatial features in the same linearized statistical space, permitting the use of linear statistics; and we adopt an optimal 3D Euclidean motion representation for more accurate rigid transformation comparisons. The method affords an efficient generative dynamic multi-object modelling platform for biological joints. We validate the method using controlled synthetic data. The shape-pose prediction results suggest that the novel concept may have utility for a range of medical image analysis applications including management of human joint disorders.