Automated Tongue Feature Extraction for ZHENG Classification in Traditional Chinese Medicine

Table 17

Comparison between using selected features versus Whole feature set for classification.

Classification experiment type	Feature selection		Whole feature
	𝐹-measure	Accuracy	𝐹-measure	Accuracy
Coating (yellow versus white)	0.764	77.10%	0.801	80.37%
ZHENG (Hot versus Cold)	0.642	65.00%	0.763	76.50%
HP Bacteria (positive versus negative)	0.636	72.38%	0.713	71.97%
Gastritis patients (superficial versus atrophic)	0.656	68.42%	0.702	71.05%
Cold ZHENG patients (superficial versus atrophic)	0.750	75.00%	0.761	76.67%
Hot ZHENG patients (superficial versus atrophic)	0.776	77.98%	0.845	84.40%
Superficial Patients (Hot versus Cold ZHENG)	0.807	80.65%	0.839	83.87%
Atrophic patients (Hot versus Cold ZHENG)	0.782	78.50%	0.734	73.83%
Normal tongue versus tongue with coating	0.833	85.88%	0.861	86.26%
Normal group versus ZHENG patients	0.834	84.68%	0.857	85.89%
Normal group versus Hot ZHENG	0.808	81.11%	0.828	82.22%
Normal group versus Cold ZHENG	0.750	75.00%	0.785	78.45%
Normal group versus superficial patients	0.765	76.52%	0.811	81.06%
Normal group versus atrophic patients	0.762	78.13%	0.837	83.85%