Table 3.

Performance comparison on the Cornell MS testing dataset (39 scans from 39 patients). Bolded and underlined numbers refer to the metrics with the best and the second-best performance, respectively.

Methods	Score	DSC	LFPR	LTPR	Precision	Sensitivity	VC	L-F1	V-F1
ALL-Net (proposed)	0.842	0.755	0.301	0.917	0.781	0.748	0.983	0.793	0.764
LST-LPA (Schmidt et al., 2012)	0.705	0.558	0.527	0.866	0.661	0.526	0.872	0.611	0.586
U-Net (Çiçek et al., 2016)	0.795	0.727	0.465	0.937	0.736	0.738	0.977	0.681	0.737
Tiramisu (Zhang et al., 2019a)	0.804	0.723	0.432	0.926	0.752	0.710	0.984	0.704	0.730
nn-Unet (Isensee et al., 2021)	0.806	0.697	0.248	0.813	0.696	0.719	0.963	0.782	0.707