Classification of SD-OCT Volumes Using Local Binary Patterns: Experimental Validation for DME Detection

Table 5

Summary of all the results in descending order.

Line

Experiment

Evaluation

Pre-processing

Feat. Detection

Mapping

Feat. Representation

Classifier

BoW

SE

SP

Type

1

2

81.2

93.7

NLM+F

LBP

✓

Local

High

SVM

✓

2

2

75.0

93.7

NLM+F+A

LBP

✓

Local

High

SVM

✓

3

2

75.0

93.7

NLM

LBP

✓

Local

High

SVM

✓

4

2

75.0

100

NLM

LBP-TOP

✓

Local

High

SVM

✓

5

2

81.2

87.5

NLM

LBP-TOP

✓

Local

High

SVM

✓

6

2

81.2

87.5

NLM+F+A

LBP-TOP

✓

Local

High

RF

✓

7

2

81.2

81.2

NLM

LBP

✓

Local

High

RF

✓

8

3

81.2

81.2

NLM

LBP-TOP

✓

Global

Low

RF

9

2

81.2

81.2

NLM+F

LBP-TOP

✓

Local

High

SVM

✓

10

3

81.2

81.2

NLM+F+A

LBP-TOP

✓

Global

Low

GB

11

3

81.2

81.2

NLM+F

LBP-TOP

✓

Global

Low

RF

12

2

75.0

87.5

NLM

LBP

✓

Local

High

-NN

✓

13

Lemaitre et al. [12]

75.0

87.5

NLM

LBP

✓

Local

High

RF

✓

14

Lemaitre et al. [12]

75.0

87.5

NLM

LBP-TOP

✓

Global

Low

RF

15

2

68.7

93.7

NLM

LBP

✓

Global

High

RF

✓

16

3

75

81.2

NLM+F+A

LBP-TOP

✓

Global

Low

RF

17

2

68.7

81.2

NLM

LBP-TOP

✓

Local

High

RF

✓

18

3

62.5

93.7

NLM

LBP-TOP

✓

Global

Low

SVM

19

3

68.7

87.5

NLM

LBP-TOP

✓

Global

Low

RF

20

3

68.7

81.2

NLM

LBP-TOP

Global

Low

RF

21

3

75.0

75.0

NLM

LBP-TOP

Global

Low

RF

22

3

68.7

75.0

NLM+F

LBP-TOP

✓

Global

Low

SVM

23

3

56.2

75.0

NLM

LBP

✓

Global

Low

RF

24

3

56.2

75.0

NLM+F

LBP

✓

Global

Low

-NN

25

3

56.2

75.0

NLM+F+A

LBP

✓

Global

Low

-NN

26

Venhuizen et al. [10]

61.5

58.8