Skip to main content
VOLUMETRIC MULTIMODALITY NEURAL NETWORK FOR BRAIN TUMOR SEGMENTATION

Abstract

Qualitative results

Figure 1 and 2. Qualitative results.

Method

Figure 3. Proposed Architecture. The kernels of the convolutions in the three pathways are 33 and no padding was made in those operations. The input of the 3 paths are centered in the same voxel, but the medium resolution and low patches are obtained from downsampled versions of the image by factors of 3 and 5, respectively

Results

METHODDICESENSITIVITYSPECIFICITYHAUSDORFF

Deepmedic
Ours		Enh. Wh. Cor.
0.69 0.86 0.68
0.71 0.88 0.68		Enh. Wh. Cor.
0.72 0.86 0.64
0.72 0.86 0.68		Enh. Wh. Cor.
0.99 0.99 0.99
0.99 0.99 0.99		Enh. Wh. Cor.
10.1 25.0 17.5
6.12 9.63 11.4
Table 1.
DATASETDICESENSITIVITYSPECIFICITYHAUSDORFF


Train
Val
Test
Enh. Wh. Cor.
0.74 0.89 0.87
0.71 0.88 0.68
0.65 0.86 0.67		Enh. Wh. Cor.
0.83 0.91 0.89
0.72 0.86 0.68		Enh. Wh. Cor.
0.99 0.99 0.99
0.99 0.99 0.99
Enh. Wh. Cor.
5.85 15.9 11.2
6.12 9.63 11.4
51.7 10.4 36.2
Table 2.

Downloads