Table 6.
Comparison of results of the proposed method and recently related works.
| Method | Dataset | Accuracy (%) | Specificity (%) | Sensitivity (%) | Dice coefficient (%) | Jaccard Index (%) |
|---|---|---|---|---|---|---|
| CNN + Morphological Optimization20 | JSRT | 98.5 | – | – | 99.2 | 98.5 |
| Atrous Convolutions14 |
JSRT MC |
– | – | – |
98.0 on JSRT 96.4 on MC |
96.1 on JSRT 94.1 on MC |
| Structured Edge Detector30 |
JSRT MC |
– | – | – |
97.6 on JSRT 95.6 on MC |
95.8 on JSRT 93.5 on MC |
| Encoder-Decoder Structure24 | JSRT | – | 99.2 | 95.2 | 96.0 | – |
| Improved FCN17 |
JSRT MC |
98.9 on JSRT 97.4 on MC |
– | – | – |
95.8 on JSRT 91.7 on MC |
| Improved SegNet18 | JSRT | 98.7 | – | – | – | 95.1 |
| U-Net21 |
JSRT MC |
97.1 on JSRT 97.7 on MC |
98.0 on JSRT 98.5 on MC |
95.1 on JSRT 95.4 on MC |
95.1 on JSRT 95.1 on MC |
– |
| AlexNetand ResNet23 | MC | 96.9 | 96.7 | 97.5 | 94.0 | 88.0 |
| Y. M. et al.31 |
JSRT MC |
– |
98.8 onJSRT 99.2 on MC |
97.9 on JSRT 98.1 onMC |
97.6 on JSRT 97.9 on MC |
95.3 on JSRT 95.9 on MC |
| Our method | JSRT | 98.5 | 98.5 | 98.4 | 97.9 | 95.8 |
| MC | 98.9 | 99.3 | 97.5 | 97.7 | 95.5 |
Significant values are in bold.