Fig. 4: Determining the accuracy of virtual double staining.

A VDS alignment of CK-stained regions in NSCLC PDEs. PDEs were stained with CK by IHC. Raw images of tumour, parenchymal and airway regions are shown as well as the corresponding VDS alignments using Platform-3. CK-stained cells/regions are highlighted in blue and stroma regions/cells highlighted in green. B Pearson’s correlation of VDS with histomorphometrist scoring of CK-positive areas within Ki67 or cPARP-stained sections. C Pearson’s correlation of VDS with histomorphometrist scoring of CK-positive areas within sections generated from vehicle or cisplatin-treated PDEs. D Pearson’s correlation of total cPARP+ve cells in tumour and stroma regions after VDS or manual segmentation of CK-positive areas. The dotted lines represent x = y + 0 (identity line). E Pearson’s correlation of total Ki67+ve cells in tumour and stroma regions after VDS or manual segmentation of CK-positive areas. The dotted lines represent x = y + 0 (identity line). F % positivity for Ki67+ve and cPARP+ve staining in tumour and stroma areas of vehicle and cisplatin-treated PDEs using either Platform-3 for VDS identification of tumour area or manual judgement. Statistical comparisons were undertaken using the Wilcoxon-paired test.