Artificial intelligence in tongue diagnosis: classification of tongue lesions and normal tongue images using deep convolutional neural network

Table 6 Accuracy, Sensitivity, Specificity and F1 Score values in each model of 5 classes

MODELS	CLASSES	ACCURACY	SENSITIVITY (%)	SPECIFICITY (%)	F1 SCORE (%)
VGG19	Coated/Hairy	83,53	84.52	97.40	84.02
	Fissured	82,10	83.80	94.59	82.93
	Geographic	85,03	81.14	94.42	83.04
	Median	84,48	79.10	98.38	82.17
	Normal	84,15	89.03	94.44	86.52
RESNET-50	Coated/Hairy	76,54	73.81	96.47	75.15
	Fissured	80,85	72.00	97.10	80.25
	Geographic	90,64	86.90	97.10	88.73
	Median	80,30	79.10	97.66	79.70
	Normal	76,02	96.13	89.96	84.90
RESNET-101	Coated/Hairy	78,89	84.52	94.47	81.61
	Fissured	81,12	81.69	94.39	81.40
	Geographic	86,36	76.00	95.31	80.85
	Median	84,21	71.64	98.38	77.42
	Normal	79,89	92.26	92.31	85.63
GOOGLENET	Coated/Hairy	81,18	82.14	97.03	81.66
	Fissured	78,47	79.58	93.56	79.02
	Geographic	84,47	77.71	94.42	80.95
	Median	79,17	85.07	97.30	82.01
	Normal	83,85	87.10	94.44	85.44
FUSION	Coated/Hairy	87,36	90.48	97.96	88.89
	Fissured	86,01	86.62	95.84	86.32
	Geographic	92,99	83.43	97.54	87.95
	Median	91,80	83.58	99.10	87.50
	Normal	86,86	98.06	95.09	92.12

ISSN: 1471-2342