An Effective Method for Detecting and Classifying Diabetic Retinopathy Lesions Based on Deep Learning

Table 7

Comparison of studies conducted with the Kaggle dataset and the proposed study.

Authors	Training type	Method	Process type	ACC	AUC	SEN
Mansour [59]	AlexNet + SVM	TL	Fundus classification	97.9	-	100
Quellec et al. [60]	CNN	EE	Fundus classification	-	95.5	-
Colas et al. [61]	CNN	EE	Fundus classification	-	94.6	96.2
Pratt et al. [62]	CNN	EE	Fundus classification	75.0	-	95.0
Jinfeng et al. [63]	CNN	TL	Fundus classification	80.3	-	-
Ours	Faster RCNN + CNN	TL	Fundus classification	99.1	99.9	99.1