Computerized image analysis as a tool to investigate the relationship between endothelial morphology and permeability

Endothelial permeability is associated with the genesis and development of atherosclerosis. Computerized image analysis is utilized to investigate the relationship between endothelial permeability and endothelial morphology. First, microscopic images are segmented to detect endothelial cells using the speckle reduction anisotropic diffusion and marker-controlled watershed, whose optimal parameter settings are obtained from the cell detection receiver operating characteristic. Two categories of morphological features are then extracted, including cell shape features and intercellular features. Finally, Student´s t-test is conducted to explore the relation of the morphology to permeability. The method correctly detected 82.3% cells in two test images, while the over-segmented and fused cells were 8.4% and 9.3%, respectively. T-tests using images from two porcine coronary arteries demonstrated that four features had significant difference (P<0.05) between regions with highest (top 25%) and lowest (bottom 25%) albumin permeability. This finding is helpful in exploring the mechanisms responsible for high permeability.