Table 2.

Comparison of different brain tumor segmentation results obtained on the BraTS2018 validation dataset.

	Dice score (DSC)[%]					95% Hausdorff distance (H95) [mm] ↓
	No. of models	WT	TC	ET	Overall	WT	TC	ET	Overall
M-UNet (17)	1	87.00	72.00	66.00	75.00	6.73	15.74	7.56	10.01
CA+DFKZ+UNet (22)	3	90.44	80.52	74.94	81.97	6.33	6.37	2.78	5.16
2D UNet (18)	3	89.1	80.9	79.3	83.1	6.99	8.96	4.12	6.69
DeepMedic+UNet (23)	26	90.5	81.3	78.8	83.53	4.32	7.56	3.81	5.17
DeeplabV3+ (19)	3	90.48	84.14	76.49	83.70	5.19	7.24	3.78	5.40
3D UNet (24)	6	90.59	83.36	78.73	84.23	4.02	5.34	3.96	4.44
DeepSCAN (20)	6	90.28	85.40	79.47	84.98	—	—	—	—
HDC-Net (25)	1	89.0	84.3	81.5	84.93	4.59	8.76	2.42	5.26
No new-Net (26)	10	90.83	85.44	81.05	85.76	4.97	7.04	2.51	4.85
The proposed	4	91.03	86.44	80.58	86.02	3.76	6.73	2.51	4.33

The results are reported as mean and are directly copied from the original papers. Bold numbers highlight the best performance in each column.

↓ indicates that a smaller value represents a better performance.