Deep Learning-Based Acute Ischemic Stroke Lesion Segmentation Method on Multimodal MR Images Using a Few Fully Labeled Subjects

Table 6

Evaluation results of the MFMF-Network whose segmentation branch is trained on different data, where “all slices” means both the normal and lesion slices are used, and “lesion slices” means that only lesion slices are used. The best result has been highlighted in italic.

	DC			F1	DR
All slices		0.702	0.931	0.801	1.000
Lesion slices	0.699 ± 0.128	0.852	0.923	0.886	1.000