Diffusion-CSPAM U-Net: A U-Net model integrated hybrid attention mechanism and diffusion model for segmentation of computed tomography images of brain metastases

Table 5 Comparison of previous studies

	Proposed Model	GAN + Mask-R-CNN + CRF [20]	PAM-U-Net [21]	ST-U-Net [21]	SEA-U-Net [21]	SERR-U-Net [21]
DSC	0.793 ± 0.133	0.726 ± 0.128	0.753 ± 0.172	0.747 ± 0.158	0.730 ± 0.135	0.718 ± 0.156
IoU	0.692 ± 0.133	0.640 ± 0.136	0.672 ± 0.159	0.667 ± 0.143	0.648 ± 0.150	0.625 ± 0.141
Accuracy	0.955 ± 0.118	0.915 ± 0.118	0.948 ± 0.125	0.930 ± 0.131	0.919 ± 0.118	0.898 ± 0.122
Sensitivity	0.803 ± 0.121	0.765 ± 0.131	0.721 ± 0.116	0.749 ± 0.120	0.702 ± 0.131	0.694 ± 0.126
Specificity	0.938 ± 0.140	0.922 ± 0.117	0.963 ± 0.104	0.951 ± 0.112	0.978 ± 0.106	0.946 ± 0.114
HD	5.606 ± 0.990	7.356 ± 0.603	6.912 ± 0.620	7.241 ± 0.835	7.539 ± 0.547	7.706 ± 0.728

DSC, Dice similarity coefficient; IoU, intersection over union; HD, Hausdorff distance

ISSN: 1748-717X