Table 6.
The comparison of proposed CardioNet and other state-of-the-art methods for the MC dataset (unit: %).
| Type | Method | Accuracy | Jaccard Index | Dice Coefficient |
|---|---|---|---|---|
| Handcrafted feature-based methods | Vajda et al. [61] | 69.0 | - | - |
| Candemir et al. [4] | - | 94.1 | 96.0 | |
| Peng et al. [55] | 97.0 | - | - | |
| Deep feature-based methods | Feature selection and Vote [62] | 83.0 | - | - |
| Feature selection with BN [62] | 77.0 | - | - | |
| Bayesian feature pyramid network [52] | - | 87.0 | 93.0 | |
| Souza et al. [30] | 96.97 | 88.07 | 96.97 | |
| HybridGNet [57] | 95.4 | |||
| MPDC DDLA U-Net [59] | 94.83 | 96.53 | ||
| CardioNet (proposed method) | 98.92 | 95.61 | 97.75 |