From: Automated diagnosis and staging of Fuchs’ endothelial cell corneal dystrophy using deep learning
Healthy Group (51 eyes of 35 patients) | FECD Group | p value | |||
---|---|---|---|---|---|
Early-stage (30 eyes of 26 patients) | Late-stage (23 eyes of 20 patients) | ||||
FECD clinical grade9 Grade: N (%) of eyes | – | 1: 14 (26%), 2: 16 (33%) | 3: 11 (17%), 4: 12 (22%) | ||
Number of eyes | Phakic eyes | 31 (60%) | 27 (51%) | P = 0.449 | |
Pseudophakic eyes (PC.IOL) | 20 (40%) | 26 (49%) | |||
Sex | Female | 24 (47%) | 33 (62%) | P = 0.078* | |
Male | 27 (53%) | 20 (38%) | |||
Age (range) years | 65 (50-82) | 69 (50-95) | P = 0.118** | ||
Mean healthy cornea prediction output | Image level | 0.99 ± 0.001 | 0.09 ± 0.012 | 0.00 ± 0.00 | P < 0.001** |
Eye level | 0.995 ± 0.003 | 0.018 ± 0.008 | 0.00 ± 0.00 | P < 0.001** | |
Mean early-stage FECD prediction output | Image level | 0.004 ± 0.001 | 0.755 ± 0.018 | 0.205 ± 0.04 | P < 0.001** |
Eye level | 0.002 ± 0.001 | 0.855 ± 0.083 | 0.033 ± 0.028 | P < 0.001** | |
Mean late-stage FECD prediction output | Image level | 0.001 ± 0.00 | 0.145 ± 0.015 | 0.795 ± 0.043 | P < 0.001** |
Eye level | 0.0001 ± 0.00 | 0.155 ± 0.087 | 0.967 ± 0.028 | P < 0.001** |