Table 4. Evaluation of models trained on the Oslo-CoMet dataset from finetuning the entire architecture.

Model	SSIM	NCC	DSC	HD	HD95	TRE	Runtime
BL-N	0.52±0.08	0.17±0.07	0.23±0.07	57.98±5.36	33.00±5.14	24.09±5.92	0.77±1.45
BL-NS	0.61±0.09	0.16±0.07	0.14±0.03	82.91±6.96	59.94±6.41	34.41±13.03	0.77±1.46
SG-ND	0.56±0.13	0.14±0.07	0.43±0.09	15.81±5.56	9.05±3.18	5.89±3.10	0.79±1.56
SG-NSD	0.58±0.13	0.14±0.07	0.42±0.10	16.26±6.37	9.50±3.51	5.84±3.01	0.76±1.48
UW-NSD	0.58±0.12	0.14±0.06	0.48±0.11	15.53±5.80	7.84±3.17	4.05±2.41	0.76±1.47
UW-NSDH	0.59±0.12	0.14±0.06	0.47±0.10	15.29±5.65	7.91±2.82	3.95±2.09	0.78±1.51
Unregistered	0.60±0.13	0.09±0.05	0.24±0.08	24.60±5.56	19.06±4.89	11.86±2.75	-

The best performing methods for each metric are highlighted in bold.