Table 8 The provide confusion matrix outlines the classification outcomes for DME, CNV, DRUSEN, and NORMAL achieved by utilizing a reinforcement-based approach along with DENSENET201 in the above experiment.

From: Reinforcement-based leveraging transfer learning for multiclass optical coherence tomography images classification

Ground truth

 

Normal

CNV

DME

Drusen

Precision

F1-(Score)

Average (specificity and recall)

Output Classes

 

Normal

248

0

0

2

1.0000

0.9903

CNV

0

247

3

0

0.9867

0.9907

DME

1

1

248

0

0.9929

0.9905

Drusen

0

4

0

246

0.9934

0.9848

Recall

0.9929

0.9912

0.9829

0.9955

  

0.9906

Specificity

1.0000

0.9826

0.9902

0.9911

0.9909

Average F1(Score) and Precision

 

0.9932

0.9890

 

Accuracy

0.9890

 
Back to article page