A multitype Fiksel interaction model for tumour immune microenvironments

Abstract

The tumour microenvironment plays a fundamental role in understanding the development and progression of cancer. This paper proposes a novel spatial point process model that accounts for inhomogeneity and interaction to flexibly model a complex database of cells in the tumour immune microenvironments of a cohort of patients with non-small-cell lung cancer whose samples have been processed using digital pathology techniques. Specifically, an inhomogeneous multitype Gibbs point process model with an associated Fiksel-type interaction function is proposed. Estimation and inference procedures are conducted through maximum pseudolikelihood, considering replicated multitype point patterns.