Figure 3

Computer Model Simulations Predict Improved Counting Accuracy for Increasing Cell Densities in 3-Color versus 1-Color Cell Labeling Strategies. (A) For a 1-color nuclear labeling approach, computer model simulations predict the number of objects counted versus cell density. (B-C) Each initial condition for the simulations consisted of seeding a rectangle (approximating a dense tissue within the embryo) with a number of cells (represented by smaller rectangles) and running each simulated image through a custom segmentation routine (Additional File 1), then calculating the number of objects. (D) For 3-color (non-mixing) and (E) 3-color (mixing) cell labeling strategies, the computer predictions are shown for number of objects counted versus cell density. The red line marks the position of cell density = 2.0. (F) The bar graph histogram shows the number of objects counted versus the count of simulations at all the different densities in the 3-color (mixing) cell labeling strategy. The percentage is binned into five percent increments. (G) Individual images of DAPI stained embryos shows the different densities within the developing chick embryo. (H) The different regions represent E1.5-E3.5 (E = embryonic day) chick embryos, with surface ectoderm, neural tube, somite, dorsal root ganglia, retinal outer layer, and limb bud densities represented in the graph. Each tissue density can be correlated with the cell densities simulated in the computer model for (A) 1-color, (D-E) and 3-color cell labeling strategies.