Table 4. Relevant high specificity cut-offs and sensitivities for best regional variables tested.
Best regional variables | AUC | 99% specificity cut-off | Sensitivity at 99% specificity | 95% specificity cut-off | Sensitivity at 95% specificity (%) |
---|---|---|---|---|---|
GSDI | 0.922* | 0.80 | 69% | 0.60 | 80.2 |
VCDR | 0.867 | 0.85 | 51.5% | 0.79 | 63.3 |
GCC FLV | 0.844 | 3.9% | 52.1% | 2.3% | 65.7 |
NFL FLV | 0.884 | 8.3% | 43.2% | 5.1% | 64.0 |
GCC GLV | 0.886 | 11.6% | 52.5% | 9.3% | 61.4 |
NFL GLV | 0.896 | 23.1% | 38.1% | 14.5% | 68.6 |
GCC overall | 0.866 | 84.0 μ | 50.0% | 85.7 μ | 56.4 |
NFL overall | 0.894 | 76.5 μ | 36.4% | 86.1 μ | 64.8 |
The AUC, sensitivity and specificity for GSDI were calculated using leave-one-out cross-validation.
AUC, area under the receiver operator characteristic curve; FLV, focal loss volume; GCC, ganglion cell complex; GLV, global loss volume; GSDI, glaucoma structural diagnostic index; NFL, nerve fibre layer; VCDR, vertical cup-to- disc ratio.