A Novel Context Aware Joint Segmentation and Classification Framework for Glaucoma Detection

Table 3

Segmentation results comparison with state-of-the-art-Drishti-gs.

Model/method	Accuracy	DC	JC	Sensitivity	Specificity	SM	EM
Modified UNet [21]; Sevastopolsky (2017)	—	0.9043	0.835	0.9156	0.9969	—	—
Ensemble CNN [23]; Zilly et al. (2017)	—	0.973	0.914	—	—	—	—
RACE-Net [24]; Chakravarty and Sivaswamy (2018)	—	0.97	—	—	—	—	—
M-Net [15]; Fu et al. (2018)	—	0.9678	0.9386	0.9711	0.9991	—	—
FC-DenseNet [49]; Al-Bander et al. (2018)	0.9969	0.949	0.9042	0.9268	0.9992	—	—
Retina- GAN (pixel GAN) [31]; Son et al. (2019)	—	0.9674	—	—	—	—	—
SAN [33]; Liu et al. (2019)	—	0.98	—	—	—	—	—
U-Shaped CNN with MSMKU [34]; Xu et al. (2019)	—	0.9780	0.9496	0.9792	0.9994	—	—
Dual UNet+MobileNetV2 [27]; Civit-Masot (2020)	—	0.93	—	—	—	—	—
CDED-Net [36]; Tabassum et al. (2020)	—	0.9597	0.9183	0.9754	0.9973	—	—
RFC-Net [38]; Gao et al. (2020)	0.9764	0.9787	—	0.9578	0.9783	—	—
Improved Image Processing Algorithms [50]; Ramani and Shanthamalar (2020)	0.9922	0.8663	—	0.9495	0.9934	—	—
FBLS [51]; Ali et al. (2020)	—	0.968	—	—	—	—	—
Attention UNet [55]; Zhao e al. (2021)	0.9975	0.9638	0.9301	0.9488	0.9993	—	—
CAE [56]; Bengani and Vadivel (2021)	0.9957	0.967	0.9314	0.9539	0.9993	—	—
GD-YNet (proposed)	0.9986	0.9945	0.9971	0.9812	0.9801	0.8271	0.8045

-Not reported.