Skip to main content
. Author manuscript; available in PMC: 2021 Jun 14.
Published in final edited form as: Lab Invest. 2020 Dec 14;101(3):396–407. doi: 10.1038/s41374-020-00511-3

Figure 2. Comparison of digital platforms.

Figure 2

a Digitisation of Ki67-stained IHC raw images using three software packages. PDEs were stained for Ki67 by IHC and visualised with DAB. The slides were then imaged using a Hamamatsu Nanozoomer XR and analysed using three digital pathology image analysis platforms (QuPath, ImmunoRatio and VisioPharm; Platforms 1-3 respectively). For all platforms, red/orange indicates DAB positivity and blue indicates DAB negative cells.

b Pearson’s correlation coefficient comparing Platform-1 and Platform-3 values for total cells counted across 56 NSCLC PDEs.

c Bland-Altman plot comparing total cell counts as determined by Platform-1 and Platform-3. The blue line indicates the mean difference between total counts, the green and red lines illustrate the upper and lower limits of agreement based on 95% CIs, respectively.

d Passing-Bablok regression model illustrating a proportional difference in total cell counts between Platform-1 and Platform-3.

e Pearson’s correlation coefficient comparing Platform-3 and Platform-1 for Ki67+ve and cPARP+ve cell counts across 56 PDEs.

f Bland-Altman plot comparing Platform-3 and Platform-1 for Ki67+ve/cPARP+ve cell counts. The blue line indicates the mean difference between positive counts while the green and red lines illustrate the upper and lower limit of agreement based on 95% CIs, respectively.

g Passing-Bablok regression model illustrates a proportional difference in positive cell counts between Platform-3 and Platform-1.