Table 5.

Performance deep learning models for wound area and wound tissue segmentation.

Feature	Architecture	Encoder	Color Calibration	Intersection over Union		Pixel Accuracy
				Mean	S.D.	Mean	S.D.
Wound area	U-Net	MobileNetV2	Yes	0.6826	0.1987	0.9881	0.0105
			No	0.6624	0.2307	0.9854	0.0219
		EfficientNet-B2	Yes	0.6964	0.1359	0.9866	0.0114
			No	0.6878	0.1665	0.9871	0.0101
Epithelialization area	U-Net	MobileNetV2	Yes	0.3186	0.1630	0.9792	0.0153
			No	0.2854	0.1552	0.9770	0.0179
		EfficientNet-B2	Yes	0.3957	0.1869	0.9808	0.0157
			No	0.3672	0.1517	0.9812	0.0145
Granulation tissue	U-Net	MobileNetV2	Yes	0.5997	0.2885	0.9899	0.0086
			No	0.5879	0.2633	0.9921	0.0048
		EfficientNet-B2	Yes	0.6421	0.2444	0.9920	0.0064
			No	0.6297	0.2529	0.9915	0.0084
Necrotic tissue	U-Net	MobileNetV2	Yes	0.1386	0.1467	0.9831	0.0133
			No	0.1185	0.1754	0.9858	0.0125
		EfficientNet-B2	Yes	0.1552	0.1428	0.9853	0.0123
			No	0.1195	0.1040	0.9896	0.0073