Table 12.

Number of parameters, training time and inference time per model after optimisation.

Model	nb parameters	Training time (h)	Inference time per image (s)
nnFormer43	$3.7\times {10}^7$	144	1.7
nnUNet33	$3.0\times {10}^7$	121	0.8
SegmentationNet44	$1.8\times {10}^7$	28	0.6
Swin-UNetr45,46	$6.2\times {10}^7$	77	2.4
Transbts48	$3.4\times {10}^7$	69	1.1
UNetr49	$9.3\times {10}^7$	17	0.7
VT-UNet50	$1.2\times {10}^7$	77	3.0