Early stages of tumor angiogenesis can be modeled by various in vitro cultures in which endothelial cells
(ECs) form networks that are considered to mimic the vascularization of tumors in vivo. Image based
quantification of EC culture model is a useful method for effective characterization of early stage in vitro
vasculogenesis and the effects of pro and anti-angiogenesis reagents. We propose an image analysis method
to quantify the EC tube formation in 2D cultures. The method segments images by high pass filtering in
Fourier space, followed by thresholding and a skeletonization and pruning process to generate the binary
skeleton image of the cell patterns in culture. Several quantities such as the network entropy (NE), the node
number, total number of chords, total and average chord length were used to quantify the evolution of EC
tubes. The automatic measurement of chord length was validated against manual measurement, achieving
an R<sup>2</sup> value of 0.953, and was used to assay for tubal extension as a function of increasing VEGF
concentration. Measurements of NE, node number, chord lengths were demonstrated on ECs network-like
patterns in culture.
Numerous investigations in the last years focused on chromosome and gene arrangements through the application of
statistical methods that analyze the non randomness of spatial distributions of fluorescence <i>in situ</i> hybridization (FISH)
labeled nucleic acid sequences in terms of their distance to the nuclear centers and their proximity to each other.
However, existing imaging processing methods are rather limited in extracting sufficient number of nuclei with FISH
label sequences, and manual analysis is unreasonably time-consuming and subjective. This paper presents an automated
system that integrates a series of advanced image processing methods to over come this rate-limiting step. Evaluation
results show that the proposed method is efficient, robust, and effective in extracting individual nuclei with FISH labels.