Myoelectric Pattern Recognition Performance Enhancement Using Nonlinear Features

Table 3

The EMG pattern recognition performance enhancement of existing feature extraction methods by the LMAV and NSV when LDA classifier is used.

	Parameter	Group	FS1	FS2	FS3	FS4
Dataset 1	Accuracy	Group 1	96.79 ± 1.00	96.27 ± 0.88	96.21 ± 0.92	96.03 ± 1.05
		Group 2	97.10 ± 0.62	96.23 ± 0.81	96.62 ± 0.86	96.54 ± 0.91
	Sensitivity	Group 1	83.96 ± 4.98	81.36 ± 4.38	81.03 ± 4.60	80.15 ± 5.25
		Group 2	85.48 ± 3.09	81.16 ± 4.05	83.10 ± 4.28	82.68 ± 4.55
	Specificity	Group 1	98.22 ± 0.55	97.93 ± 0.49	97.89 ± 0.51	97.79 ± 0.58
		Group 2	98.39 ± 0.34	97.91 ± 0.45	98.12 ± 0.48	98.08 ± 0.51
	Precision	Group 1	85.24 ± 4.72	82.41 ± 4.28	82.55 ± 4.49	81.41 ± 4.32
		Group 2	86.89 ± 3.09	82.66 ± 4.00	84.54 ± 4.20	84.00 ± 4.48
	F1 score	Group 1	83.49 ± 5.13	80.56 ± 4.57	80.58 ± 4.69	79.41 ± 5.59
		Group 2	84.94 ± 3.32	80.33 ± 4.21	82.62 ± 4.42	82.10 ± 4.75
Dataset 2	Accuracy	Group 1	97.18 ± 0.94	96.73 ± 0.93	96.69 ± 1.07	95.78 ± 1.08
		Group 2	97.87 ± 0.64	97.02 ± 0.84	97.48 ± 0.78	96.71 ± 0.86
	Sensitivity	Group 1	85.88 ± 4.72	83.66 ± 4.66	83.43 ± 5.34	78.88 ± 5.40
		Group 2	89.34 ± 3.19	85.08 ± 4.20	87.39 ± 3.91	83.53 ± 4.28
	Specificity	Group 1	98.43 ± 0.52	98.18 ± 0.52	98.16 ± 0.59	97.65 ± 0.60
		Group 2	98.82 ± 0.35	98.34 ± 0.47	98.60 ± 0.43	98.17 ± 0.48
	Precision	Group 1	87.02 ± 4.69	85.26 ± 4.31	84.79 ± 4.86	80.05 ± 4.90
		Group 2	90.49 ± 2.66	86.51 ± 3.78	88.60 ± 3.53	84.96 ± 4.07
	F1 score	Group 1	85.55 ± 4.84	83.22 ± 4.68	83.14 ± 5.34	78.37 ± 5.50
		Group 2	89.14 ± 3.19	84.76 ± 4.26	87.14 ± 3.95	83.19 ± 4.32