Table 5.
Accuracies of CardioNet and existing methods for the JSRT dataset (unit: %).
| Type | Method | Lungs | Heart | Clavicle Bone | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Acc | J | D | Acc | J | D | Acc | J | D | ||
| Local feature-based methods | Coppini et al. [53] | - | 92.7 | 95.5 | - | - | - | - | - | - |
| Jangam et al. [17] | - | 95.6 | 97.6 | - | - | - | - | - | - | |
| ASM default [54] | - | 90.3 | - | - | 79.3 | - | - | 69.0 | - | |
| Chondro et al. [25] | - | 96.3 | - | - | - | - | - | - | - | |
| Candemir et al. [15] | - | 95.4 | 96.7 | - | - | - | - | - | - | |
| Dawoud [23] | - | 94.0 | - | - | - | - | - | - | - | |
| Peng et al. [55] | 97.0 | 93.6 | 96.7 | - | - | - | - | - | - | |
| Wan Ahmed et al. [19] | 95.77 | - | - | - | - | - | - | - | - | |
| Deep feature-based methods | Dai et al. FCN [56] | - | 94.7 | 97.3 | - | 86.6 | 92.7 | - | - | - |
| Oliveira et al. FCN [35] | 95.05 | 97.45 | 89.25 | 94.24 | 75.52 | 85.90 | ||||
| OR-Skip-Net [44] | 98.92 | 96.14 | 98.02 | 98.94 | 88.8 | 94.01 | 99.70 | 83.79 | 91.07 | |
| ResNet101 [36] | 95.3 | 97.6 | 90.4 | 94.9 | 85.2 | 92.0 | ||||
| ContextNet-2 [33] | - | 96.5 | - | - | - | - | - | |||
| BFPN [52] | - | 87.0 | 93.0 | - | 82.0 | 90.0 | - | - | - | |
| InvertedNet [1] | 94.9 | 97.4 | 88.8 | 94.1 | 83.3 | 91.0 | ||||
| HybridGNet [57] | 97.43 | 93.34 | ||||||||
| RU-Net [58] | 85.57 | |||||||||
| MPDC DDLA U-Net [59] | 95.61 | 97.90 | ||||||||
|
CardioNet
(Average of Fold 1 and Fold 2) |
99.24 | 97.28 | 98.61 | 99.08 | 90.42 | 94.76 | 99.76 | 86.74 | 92.74 | |