Table 2 Classification matrices describing the performances on OCT images of human-based methods on both unprocessed (UP) and processed (P) images and CNN-based methods on both UP and P images.

From: Deep learning to distinguish Best vitelliform macular dystrophy (BVMD) from adult-onset vitelliform macular degeneration (AVMD)

OCT images

 

UP human-based

Positive

Negative

Ground truth

Positive

46

23

Negative

18

39

  

P human-based

 

Positive

Negative

Ground truth

Positive

52

17

Negative

13

44

  

UP CNN-based

Positive

Negative

Ground truth

Positive

61

8

Negative

6

51

  

P CNN-based

Positive

Negative

Ground truth

Positive

63

6

Negative

5

52

  1. CNN convolutional neural network, OCT optical coherence tomography.
Back to article page