Fig. 4

Flowchart of the simulation process. Blue, green and purple boxes contain parts of the model, model inputs and outputs, respectively. The sample grade and crypt sizes from real data input into the architecture generated. The number of cells is determined by the architecture and the user-defined cell overlap and cellularity. Cells are then iteratively generated with input of the cell phenotype distributions and the nuclear sizes and texture found in the real data. Ideal images are then degraded in order to mimic errors in an image acquisition system with parameters of noise variance defined by the user. In addition to the final image, various ground truth data is output