Table 2. Results of ablation experiments.

Backbone	CoT	MFCA	Dice (%)	Accuracy (%)	Intersection over Union (%)
nnUnet (16)	×	×	93.77	98.79	88.38
	√	×	94.42	98.81	89.69
	√	√	94.59†	98.88†	89.86†

^†, the highest performance value in the corresponding metric column. CoT, contextual Transformer; MFCA, multi-scale feature channel attention.