Fig. 5: Comparing the performances of Spatiopath and SODA for different cell sizes.
From: Statistical analysis of spatial patterns in tumor microenvironment images
a–d A cells (white masks) are randomly distributed in Ω = 1024 × 1024 pixels (1 pixel = 1 μm). Four increasing cell sizes are simulated (from left to right, μ = 1, 5, 10, and 20 μm). The positional centers of mass of the cells are represented by red dots. The spatial accumulation of points (set B) around A cells is simulated according to a Thomas process as previously described: a subset of B points is randomly distributed in Ω (blue dots), while the remaining B points are coupled to A cells (yellow dots). e–h Comparison of the association level ASI obtained from Spatiopath (blue curves, solid line = mean, shaded envelope = ±standard error, n = 10 simulations per condition) and SODA (red curves) in relation to simulated values (α, δ0). Dotted lines represent the ideal value of ASI = α. i–l Comparison of the estimated association distance (δa obtained from Spatiopath (blue curves, solid line = mean, shaded envelope = ±standard error, n = 10 simulations per condition) and SODA (red curves) in relation to simulated value δ0. Dotted lines represent the ideal value of δa = δ0. Source data are provided as a Source Data file.
