Table 8.

Effects of image preprocessing on accuracy for both datasets.

Dataset	Classification	Image Preprocessing	SVM	LR	RF
First Dataset	Abnormal vs. Normal	Yes	0.967	0.969	0.923
		No	0.956	0.95	0.77
	AMD vs. DME	Yes	0.909	0.926	0.851
		No	0.923	0.931	0.762
	CNV vs. DRUSEN	Yes	0.964	0.977	0.933
		No	0.954	0.963	0.874
Second Dataset	Abnormal vs. Normal	Yes	0.97	0.976	0.94
		No	0.986	0.983	0.896
	AMD vs. DME	Yes	0.936	0.933	0.91
		No	0.88	0.863	0.72