Table 5.
The existing DL models on TMJ and sinus disease and their functions and performance.
| Authors | DL Models |
Year | Training Dataset | Validation/Test Dataset |
Functions | Best Performance of DL | Time- Consuming |
|---|---|---|---|---|---|---|---|
| Soroushmehr et al. [65] |
U-Net | 2021 | 90 | 19 | Mandibular condyles and ramus segmentation | Sensitivity: 0.93 ± 0.06 Specificity: 0.9998 ± 0.0001 Accuracy: 0.9996 ± 0.0003 F1 score: 0.91 ± 0.03 |
No |
| Prieto et al. [66] |
Web-based system based on neural network | 2018 | 259 | 34 | TMJ OA classification |
No | No |
| Prieto et al. [67] |
SVA | 2019 | 259 | 34 | TMJ OA classification |
Accuracy: 0.92 | No |
| Ozveren et al. [68] |
CNN | 2022 | 237 | 59 | Maxillary sinusitis evaluation | Accuracy: 0.997 Sensitivity: 1 Specificity: 0.993 |
No |
| Song et al. [69] |
3D U-Net | 2021 | 70 | 20 | Sinus lesion segmentation |
DSC: 0.75~0.77 Accuracy: 0.91 |
1824 s for manual 855.9 s for DL |