TABLE III.

Summary of Refinement on Hand-Drawn Coarse Annotations

Dataset	Method		PPV	TPR	F1
	Coarse annotations		0.656 ±0.219	0.830 ±0.208	0.682 ± 0.149
CAMELYON16	DkNN		0.832 ±0.167	0.776 ±0.252	0.756 ± 0.182
	Rank Pruning		0.868 ±0.131	0.836 ±0.168	0.830 ± 0.121
	LC-MIL	LC-MIL-miNet	0.819 ±0.141	0.898 ±0.129	0.840 ± 0.096
		LC-MIL-atten	0.828 ± 0.140	0.901 ± 0.114	0.848 ± 0.096
PAIP2019	Coarse annotations		0.647 ±0.202	0.934 ±0.174	0.742 ± 0.182
	DkNN		0.753 ±0.202	0.920 ±0.179	0.808 ± 0.181
	Rank Pruning		0.799 ±0.194	0.913 ±0.155	0.830 ± 0.168
	LC-MIL	LC-MIL-miNet	0.785 ±0.202	0.932 ±0.131	0.833 ± 0.167
		LC-MIL-atten	0.805 ± 0.213	0.919 ± 0.114	0.833 ± 0.163
PAIP2020	Coarse annotations		0.705 ± 0.152	0.951 ± 0.092	0.796 ± 0.114
	DkNN		0.882 ± 0.105	0.936 ± 0.112	0.898 ± 0.095
	Rank Pruning		0.908 ± 0.091	0.853 ± 0.152	0.867 ± 0.110
	LC-MIL	LC-MIL-miNet	0.860 ± 0.112	0.897 ± 0.103	0.869 ± 0.081
		LC-MIL-atten	0.902 ± 0.097	0.844 ± 0.114	0.863 ± 0.077