Research Article
Multichannel Retinal Blood Vessel Segmentation Based on the Combination of Matched Filter and U-Net Network
Table 7
MCC and CAL metrics of existing techniques on the three datasets.
| Method | DRIVE | STARE | CHASE_DB1 | MCC | CAL | MCC | CAL | MCC | CAL |
| Azzopardi et al. (2015) [41] | 0.719 | 0.721 | 0.698 | 0.709 | 0.656 | 0.608 | Orlando et al. (2016) [42] | 0.740 | 0.675 | 0.726 | 0.665 | 0.689 | 0.571 | Dharmawan et al. (2017) [18] | 07991 | 0.8834 | 0.7959 | 0.8181 | — | — | Yang et al. (2018) [43] | 0.725 | — | 0.662 | — | — | — | Strisciuglio et al. (2019) [44] | 0.729 | 0.728 | 0.698 | 0.709 | 0.663 | 0.620 | Khan et al. (2020) [45] | 0.739 | 0.696 | 0.707 | 0.566 | 0.629 | 0.547 | 2nd human observer | 0.770 | 0.771 | 0.741 | 0.622 | 0.626 | 0.722 | Proposed method | 0.756 | 0.796 | 0.796 | 0.837 | 0.566 | 0.733 |
|
|