Learning to detect boundary information for brain image segmentation

BMC Bioinformatics

Table 5 Segmentation performance in Dice Coefficient (DC) obtained on the MRBrainS dataset achieved by our model (with and without BIM), compared to the state-of-the-art models

Model	Dice Coefficient (DC) accuracy
Model	CSF (%)	GM (%)	WM (%)
Özgün et al. [50]	83.9	88.9	89.4
Dong et al. [51]	83.5	85.4	88.9
Mahbod et al. [52]	85.5	87.3	88.7
Marijn et al. [53]	85.5	87.3	88.7
3D,FCN+MIL+G+K [17]	94.1	90.2	89.7
Multi-stage [38]	93.0	93.0	88.0
Our model (with BIM)	92.0	95.0	93.0
Our model (without BIM)	89.0	90.0	90.0

ISSN: 1471-2105