|
| Year | Image modality | Features | Method | Dataset | Accuracy |
|
| 2016 | Fundus | Optic disc, cup-to-disc ratio [17] | Wavelet feature extraction score normalization | Local database, 63 images | 94.7% |
| 2017 | Fundus | Optic disc, vessel map, vessel component [19] | Morphological operation for glaucoma detection | 10 publicly available datasets; only HRF provides center ground truth | 99.49% |
| 2016 | Fundus | CDR, texture feature, color moments, intensity [21] | Wavelet, multiwavelet binary pattern, Grey cooccurrence matrix | 50 fundus images taken from local dataset | 87% |
| 2016 | Fundus | Optic disc, CDR, regional image feature [22] | Gaussian filter bank, | RIMONE, SLO | 93.9% |
| 2020 | OCT | Disc area, cup area, CDR, RNFL thickness [23] | Hazard models along with OCT | Private | Disc area: 0.008 ( value); RNFL thickness 0.003 ( value) |
| 2017 | SDOCT | GCLIP thickness [24] | Macula by SAP, GCA, Z-test | Private: 127 eyes images of 80 participants | 75, 0.65 |
| 2016 | OCTA | RNFL thickness, capillary density [25] | Layer segmentation method | Private dataset: 67 images | 94% |
| 2018 | SDOCT | Inner macular layer thickness, inner retinal layer thickness [26] | Spectralis OCT | Private: 148 eyes of patients (Spain and Belgium Hospital) | 42.2%
88.9% |
|
