Table 4.

Confusion matrix of the classification of superpixels using the optimized Xception network in melanoma patients in 6 categories: tumor, stroma, normal epidermis, cluster of lymphocytes (Lym), fat and empty/white space (separate test set of 5 whole-slide images).

	Tumor	Stroma	Epidermis	Lym	Fat	Empty space
Tumor	5286	10	7	8	0	0
Stroma	9	986	0	0	2	0
Epidermis	22	0	545	0	1	0
Lym	0	0	1	821	0	0
Fat	0	9	0	0	5603	3
Empty space	0	0	0	0	98	681

Overall accuracy = 98.8%, average precision = 96.9%, average recall = 98.5%.

The bold values indicate the correct predictions of the network.