Table 5 Averaged dice of the propose method with different weights for segmentation network loss function. (unit: %).
From: Dual-branch image projection network for geographic atrophy segmentation in retinal OCT images
\({\lambda }_{Dice}\) | \({\lambda }_{BCE}\)=0.2 | \({\lambda }_{BCE}\)=0.5 | ||||||
|---|---|---|---|---|---|---|---|---|
\({\lambda }_{CLE}\)=0.2 | 0.5 | 1 | 2 | \({\lambda }_{CLE}\)=0.2 | 0.5 | 1 | 2 | |
0.2 | 82.54 | 84.79 | 85.34 | 82.51 | 85.83 | 86.17 | 86.49 | 82.68 |
0.5 | 84.23 | 85.32 | 86.46 | 82.63 | 86.24 | 86.89 | 87.03 | 82.91 |
1 | 84.18 | 85.96 | 86.25 | 83.86 | 86.47 | 86.64 | 86.74 | 84.36 |
2 | 83.87 | 85.28 | 85.57 | 85.23 | 86.72 | 85.72 | 86.12 | 85.57 |
\({\lambda }_{Dice}\) | \({\lambda }_{BCE}\)=1 | \({\lambda }_{BCE}\)=2 | ||||||
|---|---|---|---|---|---|---|---|---|
\({\lambda }_{CLE}\)=0.2 | 0.5 | 1 | 2 | \({\lambda }_{CLE}\)=0.2 | 0.5 | 1 | 2 | |
0.2 | 83.76 | 84.51 | 85.44 | 85.27 | 84.36 | 85.23 | 86.34 | 85.17 |
0.5 | 84.49 | 86.17 | 86.57 | 85.68 | 85.14 | 86.76 | 86.62 | 85.58 |
1 | 84.37 | 85.79 | 86.70 | 86.18 | 85.88 | 86.36 | 85.89 | 85.37 |
2 | 84.18 | 84.61 | 85.41 | 84.72 | 86.62 | 86.51 | 86.28 | 85.92 |
