Multilayer Adjusted Cluster Point Process Model: Application to Microbial Biofilm Image Data Analysis
share
A common problem in spatial statistics tackles spatial distributions of clusters of objects. Such clusters of similar or dissimilar objects are encountered in many fields, including field ecology, astronomy, and biomedical imaging. Still challenging is to quantify spatial clustering when one or more entities clusters around a different entity in multiple layers. Such multi-entity and multi-layered structures are observed, for example, in human dental plaque biofilm images, which exhibit multi-species structures in corncob-like arrangements. We propose a novel, fully Bayesian, multivariate spatial point process model to quantify corncob-like arrangements with "parent-offspring" statistical approaches. The proposed multilayer adjusted cluster point process (MACPP) model departs from commonly used approaches in that it exploits the locations of the central "parent" object in clusters and accounts for multilayered multivariate parent-offspring clustering. In simulated datasets, the MACPP outperforms the classical Neyman-Scott process model, a univariate model for modeling spatially clustered processes, by producing decisively more accurate and precise parameter estimates. We analyzed data from a human dental plaque biofilm image in which Streptococcus and Porphyromonas simultaneously cluster around Corynebacterium and Pasteurellaceae clusters around Streptococcus. The proposed MACPP model successfully captured the parent-offspring structure for all the taxa involved.
READ FULL TEXT
