Skip to main content

Table 1 Comparison between human ophthalmologists and DeepDrARN

From: Automated early detection of acute retinal necrosis from ultra-widefield color fundus photography using deep learning

Parameter

Uveitis vs. Normal

ARN vs. NAU

 

Accuracy (%)

Precision (%)

Recall (%)

F1 score

Accuracy (%)

Precision (%)

Recall (%)

F1 score

Senior ophthalmologist

90.00

95.12

92.86

93.98

90.00

94.74

92.31

93.51

Intermediate ophthalmologist 1

94.00

100.00

92.86

96.30

94.00

95.00

97.44

96.20

Intermediate ophthalmologist 2

74.00

100.00

69.05

81.69

90.00

94.74

92.31

93.51

Junior ophthalmologist 1

92.00

100.00

90.84

95.00

86.00

88.10

94.87

91.36

Junior ophthalmologist 2

92.00

97.50

92.86

95.12

76.00

96.55

71.79

82.35

Junior ophthalmologist 3

80.00

86.36

90.48

88.37

56.00

100.00

43.59

60.71

Junior ophthalmologist 4

76.00

89.47

80.95

85.00

74.00

90.63

74.36

81.69

Mean ophthalmologists

85.43

95.49

87.13

90.78

80.86

94.25

80.95

85.62

DeepDrARN

92.00

91.30

100.00

95.54

92.00

81.82

81.82

81.82

  1. ARN = acute retinal necrosis; NAU = non-ARN uveitis