Table 4. Adjusted p-values for N×N comparisons of diagnostic algorithms over 5 repetitions of 5-fold cross-validation method. The p-values below 0.05 demonstrate that the algorithms differ significantly (marked in italic font) in terms of precision values.

Hypothesis	Wilcoxon p-value	Nemenyi p-value
VGG16 vs. Extra Trees	4.07e-5	0.02638
CADNet vs. Extra Trees	1.39e-5	2.52e-5
CADNet vs. VGG16	0.00024	0.47791
CADNet + RFE.EN + GNB vs. Extra Trees	1.23e-5	1.05e-13
CADNet + RFE.EN + GNB vs. VGG16	1.39e-5	8.62e-6
CADNet + RFE.EN + GNB vs. CADNet	3.62e-5	0.01296