Automated Image Analysis of In Vitro Angiogenesis Assay

• Published in: Journal of Laboratory Automation Vol. 18; no. 5; pp. 411 - 415
• Main Authors: Boizeau, Marie-Laure, Fons, Pierre, Cousseins, Lionel, Desjobert, Josette, Sibrac, David, Michaux, Corinne, Nestor, Anne-Laure, Gautret, Bruno, Neil, Keith, Herbert, Corentin, Hérault, Jean-Pascal, Talbot, Hugues, Bono, Françoise
• Format: Journal Article
• Language: English
• Published: Los Angeles, CA SAGE Publications 01.10.2013
• Subjects: Cytological Techniques - methods; Endothelial Cells - drug effects; Humans; Image Processing, Computer-Assisted - methods; Neovascularization, Physiologic - drug effects; Software; tube formation; stereology; mathematical morphology; length measurement; validation
• ISSN: 2211-0682; 2472-6303; 1540-2452; 2211-0690
• DOI: 10.1177/2211068213495204

Angiogenesis is the biological process of generating new capillary blood vessels. It is a fundamental component of a number of normal (reproduction and wound healing) and pathological processes (diabetic retinopathy, rheumatoid arthritis, tumor growth, and metastasis). In vitro angiogenesis assays provide a platform for evaluating the effects of pro- or antiangiogenic compounds. One of the most informative assays is the endothelial cells capillary tube formation assay performed on a biological matrix. This assay is based on quantification of the stimulatory and inhibitory effects of various agents, which is estimated through the measurement of the pseudo-tubules network length. This standard measurement is usually carried out manually by trained operators but requires time, attention, and dedication to achieve a reasonable degree of accuracy. Moreover, the screening is operator dependent. In this article, we propose an automated procedure to evaluate the pseudo-tubule network lengths. We propose a series of image analysis procedures developed using a freely available image analysis software library. More than 800 images from 12 experiments were analyzed automatically and manually, and their results were compared to improve and validate the proposed image analysis procedure. The resulting image analysis software is currently running on a dedicated server, with comparable accuracy to manual measurements. Using this new automated procedure, we are able to treat 540 images, or three complete assays per hour.