[Retracted] Automated Detection and Segmentation of Exudates for the Screening of Background Retinopathy

Table 1

Performance evaluation of the proposed method at exudates-based and image-based levels.

Author	Datasets	Number of retinal fundus images used	Exudates-based assessment (%)			Image-based assessment (%)
			Sensitivity	Specificity	Accuracy	Sensitivity	Specificity	Accuracy
Reza et al. [36]	STARE; DRIVE	20	94.3%	100%	----	----	----	----
Yazid et al. [8]	STARE	15	97.8%	99%	----	----	----	----
Jaya et al. [37]	Clinical	200	94.1%	90%	93%	----	----	----
Welfer et al. [11]	DIARETDB1	89	70.4%	98.84%	----	----	----	----
Niemeijer et al. [15]	Clinical	300	----	----	----	95%	86%	----
Youssef et al. [38]	Clinical	7	----	----	----	80%	100%	----
Lahmiri et al. [39]	STARE	45	----	----	----	100%	100%	100%
Osareh et al. [14]	Clinical	142	93%	94.1%	----	95%	88.9%	----
Gracia et al. [40]	Clinical	117	88.08%	----	----	100%	83.95%	93.53%
Sanchez et al. [7]	Clinical	80	90.2%	----	----	100%	90%	----
Harangi et al. [41]	DIARETDB1	89	63%	85%	----	89%	----	----
Kaur and Mittal [35]	Multiple datasets	1307	88.85%	96.15%	93.46%	94.62%	98.64%	96.74%
Khojasteh et al. [42]	DIARETDB e-Ophtha	284	----	----	----	96%	98%	98%
Kusakunniran et al. [43]	DIARETDB1 e-Ophtha EX	284	56.4%	99.9%	99.8%	----	----	----
Balasubramanian et al. [44]	RIM-ONE	15	----	----	----	----	----	97.8%
Proposed method	Clinical	649	95.67%	98.23%	97.24%	97.45%	99.76%	98.65%
	MESSIDOR	400	91.21%	96.21%	94.12%	93.24%	97.49%	97.43%
	e-Ophtha EX	72	93.47%	94.57%	93.47%	93.87%	94.28%	93.82%
	DiaretDB1	70	92.47%	98.84%	96.78%	96.78%	97.28%	99.02%
	STARE	86	95.38%	96.98%	96.12%	94.98%	93.28%	97.98%
	DRIVE	30	98.89%	97.23%	98.23%	95.23%	95.23%	97.32%
	OVERALL	1307	93.44%	97.22%	95.68%	95.345%	98.63%	98.04%