Abstract:
|
Advanced analysis systems for pathology allow capturing spatial coordinates of all cells in immunohistochemistry images of the tumor microenvironment, but these rich spatially resolved cell data remain largely underutilized in cancer biomarker development. We consider the spatial distribution of cellular signal intensity (CSI) of protein expressions to model heterogeneity and spatial interactions in the framework of marked point processes with either continuous or categorical marks. Metrics of cellular heterogeneity of protein expression are derived as spatial indices based on conditional mean and the conditional variance of the marked point process. To quantify interactions between the tumor and immune system, we develop spatial metrics based on distributions of the nearest neighbor distances between cancer and immune cells. The utility of the new spatial metrics was investigated using various protein expressions in tissue microarrays (TMAs) incorporating tumor tissues from over 1,000 breast cancer patients.
|