Skip to main content

Table 1 Thermographic image pattern recognition: the top ten best results from a texture distance of six

From: Medical infrared thermal imaging of syringomyelia in the Cavalier King Charles Spaniel

Feature Set*

# correct (out of 93)

% correct (out of 93) - accuracy

Syrinx presence - # correct (out of 48)

Syrinx presence - % sensitivity

Syrinx absence - # correct (out of 45)

Syrinx absence - % specificity

A

65

69.89%

39

81.25%

26

57.78%

B

64

68.82%

40

83.33%

24

53.33%

C

64

68.82%

33

68.75%

31

68.89%

D

64

68.82%

32

66.67%

32

71.11%

E

64

68.82%

31

64.58%

33

73.33%

F

63

67.74%

39

81.25%

24

53.33%

G

63

67.74%

36

75.00%

27

60.00%

H

63

67.74%

35

72.92%

28

62.22%

I

63

67.74%

34

70.83%

29

64.44%

J

63

67.74%

33

68.75%

30

66.67%

  1. Feature sets included one or more of the following features: histogram mean, histogram energy, texture inertia, texture energy, texture correlation, and texture entropy