Supplementary Figure 2: Segmentation accuracy for varying levels of cell density
From: In vivo cell-cycle profiling in xenograft tumors by quantitative intravital microscopy
Cell density is estimated as the percentage of voxels belonging to the foreground/cell-nuclei in the histone channel wherein the foreground mask is computed using the locally adaptive Poisson-based minimum error thresholding method used in our nuclei segmentation algorithm. Shown below the plot are 3D volume visualizations of a few volumes in our test dataset along with their cell densities.
