Computer-Assisted Diagnosis for Diabetic Retinopathy Based on Fundus Images Using Deep Convolutional Neural Network

Table 2

Performance comparison.

Method	Year	Reference	Sn	Sp	Acc	AUC	Database
Vascular tracking	2010	[9]	0.7468	0.9551	0.9285	—	DRIVE
	2012	[10]	0.6887	0.9562	0.9290	—	STARE
Matched filtering	2009	[14]	0.6611	0.9848	0.9497	—	STARE
	2012	[15]	0.7191	0.9687	0.9407	—	STARE
			0.7154	0.9716	0.9343	—	DRIVE
Morphological processing	2011	[18]	0.7352	0.9795	0.9458	—	DRIVE
	2014	[19]	0.7862	0.9815	0.9598	—	DRIVE
Deformation model	2007	[21]	0.6634	0.9682	0.9316	—	DRIVE
	2009	[23]	—	0.9736	0.9087	—	STARE
			—	0.9772	0.9610	—	DRIVE
	2014	[24]	0.7187	0.9767	0.9509	—	STARE
			0.7354	0.9789	0.9477	—	DRIVE
Machine learning	2007	[25]	—	—	0.9584	0.9602	STARE
			—	—	0.9563	0.9558	DRIVE
	2011	[26]	0.6944	0.9819	0.9526	—	STARE
			0.7067	0.9801	0.9452	—	DRIVE
	2013	[27]	0.8194	0.9679	0.9725	—	DRIVE
	2014	[28]	0.8104	0.9791	0.9813	0.9751	STARE
			0.8173	0.9733	0.9767	0.9475	DRIVE
	2014	[29]	0.7887	0.9633	0.9441	—	STARE
			0.7512	0.9684	0.9412	—	DRIVE

Sn: sensitivity, Sp: specificity, Acc: accuracy, AUC: area under curve.