Table 6.

Evaluation metrics of our proposed model using Reinhard normalization

Folds	Confusion matrices					Performance evaluation
	Predict $\to$ Actual $\downarrow$	Ben.	Ins.	Inv.	Nor.	Prec.	Rec.	F1	Test	Accuracy (%)	Kappa
Fold 1	Benign	43	4	1	2	1.00	0.86	0.92	50	96.44	0.945
	In situ	0	49	1	0	0.89	0.98	0.93	50
	Invasive	0	1	225	4	0.98	0.98	0.98	230
	Normal	0	1	2	117	0.95	0.97	0.96	120
Fold 2	Benign	46	2	0	2	0.98	0.92	0.95	50	96.88	0.952
	In situ	1	49	0	0	0.89	0.98	0.93	50
	Invasive	0	3	223	4	1.00	0.97	0.98	230
	Normal	0	1	1	118	0.95	0.98	0.97	120
Fold 3	Benign	47	2	1	0	0.98	0.94	0.96	50	96.44	0.944
	In situ	1	48	1	0	0.92	0.96	0.94	50
	Invasive	0	1	226	3	0.97	0.98	0.97	230
	Normal	0	1	6	113	0.97	0.94	0.96	120
Fold 4	Benign	47	2	0	1	0.96	0.94	0.95	50	97.11	0.955
	In situ	1	47	1	1	0.94	0.94	0.94	50
	Invasive	0	0	227	3	0.99	0.99	0.99	230
	Normal	1	1	2	116	0.96	0.97	0.96	120
Fold 5	Benign	47	3	0	0	0.87	0.94	0.90	50	95.33	0.928
	In situ	2	47	0	1	0.87	0.94	0.90	50
	Invasive	2	2	223	3	0.99	0.97	0.98	230
	Normal	3	2	3	112	0.97	0.93	0.95	120
Final	Benign	47	2	0	1	0.98	0.94	0.96	50	97.33	0.959
	In situ	1	48	1	0	0.92	0.96	0.94	50
	Invasive	0	1	226	3	0.99	0.98	0.98	230
	Normal	0	1	2	117	0.97	0.97	0.97	120