Skip to main content
. 2015 Nov 5;10(11):e0142309. doi: 10.1371/journal.pone.0142309

Table 2. Intra Class Correlation coefficients for all participants (n = 46).

Observer reliability measure The manual protocol ICC and 95%CI* The semi-automated protocol ICC and 95%CI Comparison of semi-auto to manual P-Value The fully automated protocol ICC and 95%CI Comparison of fully auto to manual P-Value Comparison of fully auto to semi-auto P-Value
CNFL Inter 0.73(0.55,0.84) 0.75(0.58,0.86) 0.39 0.78(0.34,0.91) 0.40 0.46
Intra 0.72(0.50,0.85) 0.73(0.56,0.84) 0.43 0.84(0.73,0.91) 0.021 0.039
CNFD Inter 0.61(0.39,0.76) 0.68(0.48,0.81) 0.68 0.50(0.17,0.71) 0.83 0.95
Intra 0.57(0.33,0.73) 0.60(0.39,0.76) 0.37 0.63(0.41,0.78) 0.27 0.38
CNBD Inter 0.32(-0.07,0.61) 0.40(0.13,0.61) 0.28 0.50(0.23,0.69) 0.093 0.23
Intra 0.61(0.40,0.77) 0.31(0.02,0.55) 0.99 0.43(0.16,0.64) 0.95 0.18

ICC: Intra-class correlation coefficient (2, 1). Inter refers to inter-observer reproducibility, intra to intra-observer reproducibility. The manual protocol included manual image selection and manual image analysis. The semi-automated protocol included manual image selection and automated analysis. The fully-automated protocol included automated image selection and automated analysis. CNFL, corneal nerve fibre length. CNFD, corneal nerve fibre density. CNBD, corneal nerve branch density.

* Manual protocol ICC have been presented in a previous publication and are shown here for comparison.

† For the 11-member sub-population with CNFL below 12.3mm/mm2, the inter-observer ICC for CNFLManual, CNFLSemi-Automated, and CNFLFully-Automated were 0.54(-0.09,0.85), 0.51(-0.04,0.83), and 0.73(0.25,0.92), respectively. P-values for comparison of the ICC of CNFLSemi-Automated to CNFLManual, CNFLFully-Automated to CNFLManual, and CNFLFully-Automated to CNFLSemi-Automated were 0.56, 0.17, and 0.15, respectively. The corresponding intra-observer ICC for CNFLManual, CNFLSemi-Automated, and CNFLFully-Automated were 0.74(0.26,0.92), 0.68(0.20,0.90), and 0.86(0.55,0.96), respectively. The p-values for the corresponding comparisons were 0.64, 0.16, and 0.19.