Table 1. Performance determined from retina and vascular image of DSWC model datasets.
| Method | Dynamic threshold algorithm | U-Net model (deep learning) |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Train | Test | Train | Test | Train | Test | Train | Test | ||
|
| |||||||||
| Image type | Retinal image | Retinal image | Retinal image | Vascular image of DSWC model | Vascular image of DSWC model | Vascular image of DSWC model and Retina image | Vascular image of DSWC model | ||
| Accuracy (%) | 94.75 | 95.59 | 85.12 | 90.17 | 90.64 | ||||
| Sensitivity (%) | 62.78 | 74.90 | 14.23 | 78.98 | 80.12 | ||||
| Specificity (%) | 97.69 | 98.57 | 97.36 | 92.66 | 92.83 | ||||