This article is part of the supplement: Proceedings of the Sixth Annual MCBIOS Conference. Transformational Bioinformatics: Delivering Value from Genomes
Automatic identification of angiogenesis in double stained images of liver tissue
1 Information Technology Research, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA
2 Department of Pathology, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA
3 Biostatistics, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA
BMC Bioinformatics 2009, 10(Suppl 11):S13 doi:10.1186/1471-2105-10-S11-S13Published: 8 October 2009
To grow beyond certain size and reach oxygen and other essential nutrients, solid tumors trigger angiogenesis (neovascularization) by secreting various growth factors. Based on this fact, several researches proposed that density of newly formed vessels correlate with tumor malignancy. Vessel density is known as a true prognostic indicator for several types of cancer. However, automated quantification of angiogenesis is still in its primitive stage, and deserves more intelligent methods by taking advantages accruing from novel computer algorithms.
The newly introduced characteristics of subimages performed well in identification of region-of-angiogenesis. The proposed technique was tested on 522 samples collected from two high-resolution tissues. Having 0.90 overall f-measure, the results obtained with Support Vector Machines show significant agreement between automated framework and manual assessment of microvessels.
This study introduces a new framework to identify angiogenesis to measure microvessel density (MVD) in digitalized images of liver cancer tissues. The objective is to recognize all subimages having new vessel formations. In addition to region based characteristics, a set of morphological features are proposed to differentiate positive and negative incidences.