Automated image analysis of HUVEC morphology and Sema3E-induced cell collapse. (A) Raw images of cells acquired from a well of a screening plate by automated microscopy. DiI labeling combined with image analysis algorithms enabled automated quantification of cellular morphological changes at the screening end point. (B) Cellular membrane boundaries were overlaid on top of the raw image so that cell surface area could be automatically measured. (C) Cell protrusions were labeled in blue on top of the segmented cell masks so that the number of protrusions per cell could be counted. The green regions were the bumps on the cell surface that were too short to be classified as protrusions. Bar, 100 µm. (D) Physical parameters used to classify collapsed cells were determined based on the cell surface area distribution from the control wells (top, histogram) in combination with cell protrusion counts. Cells were classified as collapsed based on a reduction in their surface area and/or the presence of protrusions. (E) siRNAs that blocked Sema3E-induced HUVEC collapse were classified as strong hits if the percentage of cells collapsed was 3 SDs away from the plate-matched negative controls and within 3 SDs of the plate-matched positive controls (PlexinD1 siRNA; red box). siRNAs that partially blocked collapse were classified as weak hits (blue box).