Table 7. Comparison of results between proposed method and the state-of-the-art methods.

Methods	Training/Test	Magnification	Image_accuracy(%)	Patient_accuracy(%)
Gour et al. [45]	70/30 (protocol)	40×	87.40±3.00	87.47±3.22
		100×	87.26±3.54	88.15±2.97
		200×	91.15±2.30	92.52±2.84
		400×	86.27±2.18	87.78±2.46
Alkassar et al. [46]	70/30 (protocol)	40×	99
		100×	98.5
		200×	98.5
		400×	98
Li et al. [31]	50/20/30	40×	89.5±2.0	89.1±3.6
		100×	87.5±2.9	85.0±5.1
		200×	90.0±5.3	87.0±6.0
		400×	84.0±2.9	84.5±3.6
Sharma et al. [47]	80/20	40×	89.31
		100×	85.75
		200×	83.95
		400×	84.33
Celik et al. [48]	80/20	Magnification independent	99.11	89.88
Yari et al. [39]	6011/1142/406	40×	100
		100×	100
		200×	98.08
		400×	98.99
		Magnification independent	99.26
Liu et al. [40]	Random 5 folds	40×	99.33
		100×	99.04
		200×	98.84
		400×	98.53
		Magnification independent	99.24
Budak et al. [49]	Random 5 folds	40×	95.69±1.78
		100×	93.61±2.28
		200×	96.32±0.51
		400×	94.29±1.86
Mewada et al. [50]	Random 70/30	40×	97.58
		100×	97.44
		200×	97.28
		400×	97.02
Nahid et al. [51]	Didn’t mention	40×	90
		100×	85
		200×	90
		400×	91
Our method	70/30	40×	96.75±1.96	96.33±2.14
		100×	95.21±2.18	95.26±2.60
		200×	96.57±1.82	96.09±1.79
		400×	93.15±2.30	92.99±2.85
		Magnification independent	95.56±2.14	95.54±2.40