Fully semantic segmentation for rectal cancer based on post-nCRT MRl modality and deep learning framework

BMC Cancer

Table 2 Evaluation metrics (DSC, Jaccard, Precision, Recall, and F1-score) with fivefold cross-validation

Fold	DSC	Jaccard	Precision	Recall	F1-score
0	0.878 (0.871–0.886)	0.794 (0.784–0.805)	0.875 (0.865–0.885)	0.898 (0.891–0.906)	0.878 (0.871–0.886)
1	0.885 (0.878–0.893)	0.804 (0.795–0.814)	0.886 (0.876–0.895)	0.900 (0.894–0.906)	0.885 (0.878–0.893)
2	0.882 (0.875–0.889)	0.798 (0.789–0.808)	0.882 (0.873–0.891)	0.897 (0.890–0.904)	0.882 (0.875–0.889)
3	0.879 (0.871–0.887)	0.796 (0.786–0.806)	0.878 (0.868–0.887)	0.899 (0.891–0.906)	0.879 (0.871–0.887)
4	0.881 (0.874–0.889)	0.799 (0.789–0.809)	0.880 (0.870–0.890)	0.899 (0.892–0.906)	0.881 (0.874–0.889)
Mean	0.881 (0.879–0.884)	0.798 (0.795–0.802)	0.880 (0.876–0.884)	0.899 (0.898–0.900)	0.881 (0.879–0.884)

ISSN: 1471-2407