Table 3 Results of three statistical tests to investigate the role of specific anatomical structures in the predictions. * and ** mark significant and highly significant p-values (\(\alpha = 0.05\) and \(\alpha = 0.01\)). Bonferroni correction is applied to the Wilcoxon signed-rank tests (\(\alpha _{{\rm ind}}=0.00714\) and \(\alpha _{{\rm ind}}=0.00142\)). The tests compare patches containing anatomical structures with ones that show background.
Tested structure(s) | N (per group) | p-value | |
|---|---|---|---|
Friedman test | All classes | 135 | \(4.780\times 10^{-09}\)** |
Wilcoxon signed-rank test | Sup. temporal arcade | 135 | \(1.292\times 10^{-09}\)** |
Sup. nasal arcade | 135 | \(1.058\times 10^{-03}\)** | |
Inf. temporal arcade | 135 | \(2.080\times 10^{-09}\)** | |
Inf. nasal arcade | 135 | \(2.624\times 10^{-02}\) | |
Macula | 135 | \(1.684\times 10^{-01}\) | |
Optic disc | 135 | \(5.390\times 10^{-07}\)** | |
Wilcoxon signed-rank test | Arteries vs. veins | 135 | \(1.416\times 10^{-02}\) |
