Table 2.

Experimental results on BraTs 2019 validation leaderboard. Our team “MLKCNN-COMSATS-MIDL” results are available on portal

	Team	Architecture type	Dice_ ET	Dice_ WT	Dice_ TC	Sensitivit y_ET	Sensitivity_WT	Sensitivit y_TC	Hausdorf f_ET	Hausdorff_WT	Hausdorf f_TC
Single-model method	Proposed MLKCNN	ML-KCNN	0.73	0.88	0.81	0.77	0.89	0.79	6.47	8.52	7.45
	Proposed MLKCNN + Post- Processing	ML-KCNN	0.74	0.90	0.83	0.79	0.92	0.83	3.74	4.89	5.75
	Vu Hoang Minh et al. [42]	3D Attention U- Net	0.70	0.89	0.79	0.75	0.90	0.81	7.05	6.29	8.76
	Mehdi Amian et al. [39]	Two-way U-net architecture	0.71	0.86	0.76	0.68	0.84	0.75	6.91	8.42	11.549
	Mohammad Hamghalam et al. [37]	2D-Unet-based generative adverserial architecture	0.76	0.89	0.79	0.76	0.89	0.77	4.64	6.99	8.43
	Rupal R. Agravat et al. [38]	Proposed dense module based U-Net	0.59	0.73	0.65	0.59	0.67	0.64	9.02	16.70	16.69
Ensemble-based method	McKinley et al. [43]	Ensemble of 3D-to-2D CNNs	0.77	0.91	0.83	-	-	-	3.92	4.52	6.27
	Vu [42]	TuNet:end-to-end hierarchical brain tumor segmentation using cascaded networks	77.38	90.34	79.14	-	-	-	4.29	8.80	3.57
	Murugesan et al. [40]	Multidimensional and multiresolutional ensemble	0.77	0.89	0.78	-	-	-
	Jiang et al. [41]	Cascade ensemble of 12 models	0.80	0.90	0.86	-	-	-	3.14	5.4	4.2