A Novel Brain Image Segmentation Method Using an Improved 3D U-Net Model

Table 2

Image segmentation results of each comparison model.

Index	Segmented organization	CNN	U-net	3D U-net	Proposed
Dice	WT	0.7992	0.8285	0.8428	0.8787
	TC	0.7844	0.8006	0.8211	0.8456
	ET	0.7020	0.7602	0.7993	0.8004
	Mean	0.7619	0.7964	0.8211	0.8416
PPV	WT	0.8002	0.8473	0.8687	0.8992
	TC	0.7946	0.8331	0.8505	0.8668
	ET	0.7553	0.7717	0.7899	0.8321
	Mean	0.7837	0.8174	0.8364	0.866
Sensitivity	WT	0.8335	0.8586	0.8640	0.8899
	TC	0.8664	0.8903	0.8947	0.9078
	ET	0.7608	0.8192	0.8193	0.8335
	Mean	0.8202	0.856	0.8593	0.8771
Hausdorff	WT	2.9746	2.5762	2.6227	2.5002
	TC	1.8345	1.7481	1.7008	1.5101
	ET	3.2023	2.8350	2.8088	2.6443
	Mean	2.6704	2.3864	2.3774	2.2182