Table 3.
Comparison of the Rate of Success in Locating the ONH in Images of the Retina Obtained by Several Methods
| Method of detection | DRIVE (%) | STARE (%) | Other database (%) |
|---|---|---|---|
| Lalonde et al.21 | – | 71.6a | 100 |
| Sinthanayothin et al.27 | – | 42a | 99.1 |
| Osareh et al.14 | – | 58a | 90.3 |
| Hoover and Goldbaum16 | – | 89 | – |
| Foracchia et al.30 | – | 97.5 | – |
| ter Haar20 | – | 93.8 | – |
| Park et al.6 | 90.3 | – | – |
| Ying et al.32 | 97.5 | – | – |
| Kim et al.31 | – | – | 91 |
| Fleming et al.9 | – | – | 98.4 |
| Park et al.33 | – | 91.1 | – |
| Youssif et al.1 | 100 | 98.8 | – |
| Zhu et al.3,25 | 90 | 44.4 | – |
| Present work | 100 | 69.1 | – |
In the present work, for the images of the DRIVE database, a result is considered to be successful if the detected ONH center is positioned within 46 pixels of the manually identified center. For the STARE database, the threshold is 60 pixels
aValues are from ter Haar20