Research Article

A Comprehensive Convolutional Neural Network Survey to Detect Glaucoma Disease

Table 3

Review of glaucoma diagnosis using CNN.

YearAlgorithmFeatureMethodDatasetAccuracy

2016Deep CNNOptic disc segmentationCNN using 5 layers with VGG-19 networkDRIONS, RimOneR3F-score: 96
2017CNNOptic cup and disc segmentationSampling methodologyDrishti-GS1Accuracy: 94.10%
2017CNNDisc segmentationCNN with 7 layersDriveAccuracy: -Sensitivity: 87
2017Ant colony optimizationOptic cup segmentationAnt colony optimizationRimoneR3Accuracy: -AUC: 0.7957
2017DCNNOptic cup and disc segmentationU-Net convolutional neural networkDrishti-GS2017
2018CNNOptic disc segmentationTwo-branch CNNPrivateAccuracy: 81.69
2018Ensemble learning CNNOptic disc segmentationCNN using CDR ratioRimoneAccuracy: 81.69
2018DCNNFundus photographyCross Entropy Inception V3Private (Kin’s Eye Hospital, South KoreaAccuracy: 87.9
2018CNNONH fundus imagesGON in fundus photographsPrivate (the ADAGES new study and Alabama, California)Accuracy: -AUC: 0.91
2018DCNNOptic cup and disc segmentationMorphometric features for CNN multistage modelDrishti-GS1 RimoneAccuracy: 88.9%
2018CNNOptic disc segmentationEntropy sampling, Ensemble learningDrishti-GSAccuracy: 91.9%
2019Joint RCNNOptic cup and disc segmentationGenerative adversarial network GL-NetDrishti-GS1 OrigaAccuracy: -AUC: 90.10
2019CNNProposed deep CNNCNN with 18 layersPrivate Kasturba medical college IndiaAccuracy: 98%
2020CNNOptic cup and disc segmentationCNN OverFeat and VGG-SPrivateAccuracy: 90.50%
2020DCNNOptic Cup and Disc segmentationJoint RCNNOrigaAccuracy: 92.50%