Quantitative analysis of vascular structures geometry using neural networks

Abstract

Vascularization is defined as the sprouting of new blood vessels by expansion of the endothelium by proliferation, migration and remodeling. Vascularization is fundamental to healing, reproduction as well as embryonic development. It also plays a key role in tumor growth, tumor metastasis and other pathological processes. Understanding biological phenomena driving the creation of vascular structures is therefore essential for clinical treatment of cancer and other vascularization-related diseases. Recently, an analytical model capable of mimicking the process of in-vitro vascular network creation from randomly seeded endothelial cells has also been proposed. This paper presents the development of a novel neural network based segmentation technique working on phase contrast microscopy snap photographs of cultured endothelial cells which allows for cell structures geometry quantitative analysis thus constituting a key instrument in the development of computerized tools for vascularization parameters measurement as well as supporting also analytical model deployment and validation.