Investigations of CNN for Medical Image Analysis for Illness Prediction

Table 9

Using digital fundus images of diabetic retinopathy, the DCNN can determine whether or not there are any malign qualities in the soft and hard exudates.

Observed		Cumulative
Malignity	True	False	True	False	FPR	TPR	AUC
			0	0	1.000000	1.000000	0.082437
1	46	6	46	6	0.917563	0.981013	0.126582
2	72	8	118	14	0.788530	0.955696	0.157570
3	92	12	210	26	0.623656	0.917722	0.177624
4	108	15	318	41	0.430108	0.870253	0.185592
5	119	26	437	67	0.216846	0.787975	0.132741
6	94	68	531	135	0.048387	0.572785	0.013344
7	13	78	544	213	0.025090	0.325949	0.004673
8	8	52	552	265	0.010753	0.161392	0.001735
9	6	32	558	297	0.000000	0.060127	0.000000
10	0	19	558	316	0.000000	0.000000	0.000000
	558	316					0.882299