Table 2.

Confusion Matrix Without Normalization

		Predicted Class
Image Category	Actual Class	K	N	S	Total
Four-map selectable display	K	789	8	33	830
	N	0	887	0	887
	S	0	0	857	857
Front elevation map	K	764	0	66	830
	N	0	887	0	887
	S	34	0	823	857
Back elevation map	K	796	18	16	830
	N	0	869	18	887
	S	26	8	823	857
Pachymetry map	K	830	0	0	830
	N	0	878	9	887
	S	0	9	848	857
Front sagittal curvature map	K	813	0	17	830
	N	0	887	0	887
	S	9	26	822	857

The numbers represent the average of 10 training/validation trials by the CNN using each image category and random seeds.