Skip to main content

Table 1 Performance of the artificial neural network compared with statistical methods with regard to classifying fixation as central versus paracentral fixation

From: Detecting central fixation by means of artificial neural networks in a pediatric vision screener using retinal birefringence scanning

 

Neural network

Simple threshold

Discriminant analysis

2D

3D

4D

CAL

SBJ

CAL

SBJ

CAL

SBJ

CAL

SBJ

CAL

SBJ

SNS

1.000

0.9851

0.9917

0.9701

0.9083

0.8955

0.9417

0.8358

0.9417

0.8507

SPC

1.000

1.0000

0.9625

1.0000

0.9771

1.0000

1.0000

1.0000

1.0000

1.0000

  1. Data: CAL calibration data obtained from 600 controlled measurements from 5 test subjects (12 measurements on both eyes for each target location), SBJ clinical data from 39 subjects (78 eyes)
  2. Stats: SNS Sensitivity = True Positive/(True Positive + False Negative) = TP/(TP + FN), SPC Specificity = True Negative/(True Negative + False Positive) = TN/(TN + FP)
Back to article page

BioMedical Engineering OnLine

ISSN: 1475-925X

Contact us