Spatial analysis for the distribution of cells in tissue sections

Abstract: Spatial analysis, playing an essential role in data mining, is applied in a considerable number of fields. It is because of its broad applicability that dealing with the interdisciplinary issues is becoming more prevalent. It aims at exploring the underlying patterns of the data. In this project, we will employ the methodology that we utilize to tackle spatial problems to investigate how the cells distribute in the infected tissue sections and if there are clusters existing among the cells. The cells that are neighboring to the viruses are of interest. The data were provided by the Medetect Company in the form of 2-dimensional point data. We firstly adopted two common spatial analysis methods, clustering methods and proximity methods. In addition, a method for constructing a 2-dimensional hull was developed in order to delineate the compartments in tissue sections. A binomial test was conducted to evaluate the results. It is detectable that the clusters do exist among cells. The immune cells would accumulate around the viruses. We also found different patterns near and far away from viruses. This study implicates that the cells are interactive with each other and thus present the spatial patterns. However, our analyses are restricted in a planar circumstance instead of treating them in 3-dimensional space. For the further study, the spatial analysis could be carried out in three dimensions.