Figure 2.

Endothelial surface modelling using topographical method enhances data extraction from high-throughput OOC experiments. (A) The topographical surface method is applied to an OOC seeded with HUVECs and imaged on day 0. A surface is generated by fitting a thin plate spline to positional endothelial data and is visualized in 2-D, with x and y position shown on the x and y axes and z-height shown using a gradual color scale. The surface demonstrates wave-like contours and a gradual decline in z-height from high x positions to low x positions. (B) The topographical surface method from A is compared to the use of a flat surface to quantify tumor invasion. In the flat line method (left), objects that are below the mean z-height plus standard error of the endothelial centroids are labelled as invading. The topographical method (right) labels objects as invading when they are below the endothelial surface plus standard error. In either method, objects labelled as invading are denoted in blue. Objects not labelled as invading are denoted in orange. (C) The topographical surface workflow was built into a publicly available web tool requiring no computational background to use. The user interface of the tool (left) provides options to upload data and visualize results in a table or graph form. The tool includes options to plot invaded tumor cell objects by their x and y position (right upper) and x and z position (right lower) for quality control. These plots captured invaded objects on the edge of the OOC model missed in image analysis quality control, indicated with arrows.