An Efficient Retinal Segmentation-Based Deep Learning Framework for Disease Prediction

Table 4

Comparison results of existing with proposed method.

Existing methods	Alom [4]						Wang [18]		Jiang [11]			Guo [34]			Proposed
Methods	RU-Net			R2U-Net			Dense U-net		FCN			BSCN			EDLFDPRS
Dataset	D	S	C	D	S	C	D	S	D	S	C	D	S	C	D	S	C
ACC	95.56	97.06	96.22	95.56	97.12	96.34	95.11	95.38	96.46	96.33	97.7	98.46	98.72	98.89	97.32	97.33	97.44
score	91.55	93.96	78.1	91.71	84.75	92.8	—	—	76.75	77.55	82	82.36	85.47	75.6	83.85	92.3	81.94
Sensitivity	77.51	81.08	74.59	77.92	82.98	77.56	79.86	79.14	66.63	83.32	83.23	81.79	87.51	79.72	82.56	82.12	83.92
Specificity	98.16	98.71	98.36	98.13	98.62	98.2	97.36	97.22	99.33	97.4	98.67	98.79	98.94	98.96	98.68	98.57	98.45
AUC	97.82	99.09	98.03	97.84	99.14	98.15	97.4	97.04	97.8	97.9	99.12	98.74	99.41	98.74	98.79	98.84	98.79

Note: D: DRIVE; S: STARE; C: CHASE; ACC: accuracy; sensitivity; specificity; AUC: area under the ROC curve.