Table 3.

Analysis of the proposed CPA decoder with existing decoder architectures used for semantic segmentation on datasets described in Section 4.1. To ensure fair comparison the models were trained and tested on the same data partitions.

S.No.	Method	Jun Ma dataset [30]		MosMedData [31]		Combined dataset
		DSC	IoU	DSC	IoU	DSC	IoU
1	Point-wise attention decoder [32]	75.19	64.35	72.77	62.08	73.77	62.98
2	Stride spatial pyramid pooling and dual attention decoder [23]	79.33	67.29	74.14	65.24	76.25	67.85
3	Cross-granular attention decoder [33]	78.13	69.81	83.85	73.19	78.89	68.65
4	Proposed CPA decoder	82.63	72.20	83.49	72.78	80.12	70.42