Development of a single cell segmentation algorithm for tumour tissues.

<p>Confocal images of agar pellets of EWS-FLI1 positive Ewing sarcoma cell lines were used to optimise image segmentation. <b>a</b>. Multi-channel and single channel images (with segmentation lines) of the CHP-100 cell line in cores labelled with DAPI, CD99 and EGR1 biomarkers indicating nucleus and cytoplasm localisation, respectively (see high magnification insert). In <b>b</b>., >500 cells were manually segmented and compared to the image segmentation algorithm (see Fig. S1 in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0107105#pone.0107105.s004" target="_blank">File S1</a>) using Bland-Altman and Hausdorff distance. Also, example distributions are shown for nuclear DNA content and nuclear and cytoplasmic localisation of EGR1. Image segmentation was applied to tissue microarrays of <b>c</b>. Ewing sarcoma core biopsies on a tissue microarray (TMA), and <b>d</b>., multi-channel confocal images captured for DAPI, CD99 and Ki67 proliferation marker. In <b>e</b>. >500 cells from TMA cores were manually segmented and compared to the image segmentation algorithm using Bland-Altman and Hausdorff distance, with distributions for DNA and Ki67. Bars: 20 µm (<b>a</b>. and <b>d</b>.), and in <b>c</b>, 200, 100 and 50 µm, respectively.