Figure 1

Automated quantification of survivin immunohistochemistry. (A) Representative areas from tumour positive, tumour negative, tumour stroma, tumour stroma cells and slide background were selected as a training set to develop a pattern recognition algorithm. (B) Examples of IHC (20X) showing nuclear and cytoplasmic expression of survivin and mark-up images with the corresponding output from the automated quantification approach. The algorithm maps survivin expression by labelling nuclear expression in red and cytoplasmic expression in orange and yellow. (C) Box-plot demonstrating relationship between survivin nuclear and cytoplasmic autoscores. (D) Bar chart showing normalised prognostic importance of SNAS, CNR and survivin cytoplasmic autoscore, as defined by the decision tree classification and regression trees.