Research Article
DWT-Net: Seizure Detection System with Structured EEG Montage and Multiple Feature Extractor in Convolution Neural Network
Table 5
Results of the proposed seizure detection system in comparison with results reported in other work.
| | System/ value | Sensitivity | Specificity | FAR |
| Deep learning methods in [14] | HMM | 30.32% | 80.07% | 244 | HMM/SdA | 35.35% | 73.35% | 77 | HMM/LSTM | 30.05% | 80.53% | 60 | IPCA/LSTM | 32.97% | 77.57% | 73 | CNN/MLP | 39.09% | 76.84% | 77 | CNN/LSTM | 30.83% | 96.86% | 7 | Our DWT-Net system based on 5-second DWT-Net | | 25.44% | 97.05% | 3 | = 76% | 30.25% | 96.64% | 4 | | 33.45% | 96.10% | 5 | | 42.35% | 94.93% | 6 | = 35% | 59.07% | 89.72% | 12 | Our DWT-Net system based on 1-second DWT-Net | | 19.40% | 95.01% | 6 | = 83.2% | 30.07% | 93.83% | 7 | | 33.45% | 96.10% | 5 | | 42.35% | 94.93% | 6 | = 62.6% | 49.29% | 72% | 12 |
|
|