Quantification of angiogenesis in breast cancer by automated vessel identification in CD34 immunohistochemical sections
Mikalsen, L.Tore.Gyland.; Dhakal, H.Prasad.; Bruland, O.S.; Nesland, J.M.; Olsen, D.Rune.
Anticancer Research 31(12): 4053-4060
2011
ISSN/ISBN: 0250-7005 PMID: 22199261 Document Number: 650755
Tumor growth is dependent upon angiogenesis. Tumor vascularity, as measured by microvessel density or Chalkley counts, has been shown to predict treatment outcome. However, many issues related to reproducibility and methodology have prevented its clinical application. We present a method of automatic vessel identification applied to CD34 immunohistochemical sections which facilitates increased reproducibility. Pixel colour information was used to identify CD34 stain. In order to reduce the effects of noise and background, stained areas smaller than 3.5 μm were ignored. Comparing automatic and manual vessel counts in 50 randomly selected breast cancer cases, the method achieved an intraclass correlation coefficient of r(a)(2)=0.96 and a 95% confidence interval for the percentage difference between the counts from -26.1% to 10.8%. The method was also found to have a sensitivity approaching 100%. The method can reliably be used on colour photographs of staining for CD34 to quantify angiogenesis.