Improving Minimum Cross-Entropy Thresholding for Segmentation of Infected Foregrounds in Medical Images Based on Mean Filters Approaches

Table 2

Average evaluation values of supervised performance measurement using the proposed mean estimation techniques compared with the classical mean or Gaussian distribution for MCET.

Estimation			Skin lesion	MRI Brain
1	Classical mean		0.8923	0.8210
2	Lognormal [6]		0.8966	0.8349
3	Harmonic		0.8903	0.8323
4	Geometric		0.8593	0.8233
5	Contraharmonic	Q = −3.0	0.8830	0.8513
6		Q = −1.0	0.8903	0.8313
7		Q = −0.5	0.8938	0.8475
8		Q = 0.0	0.8923	0.8240
9		Q = 0.1	0.8875	0.8421
10	Alpha trim	d/2 = 10.0	0.8957	0.8670
11		d/2 = 20.0	0.9060	0.8922
12		d/2 = 30.0	0.9059	0.9160
13		d/2 = 40.0	0.9093	0.8988
14		d/2 = 50.0	0.9102	0.8860
15		d/2 = 55.0	0.9208	0.8839
16		d/2 = 60.0	0.9097	0.8772
17		d/2 = 65.0	0.9013	0.8781
18		d/2 = 70.0	0.8849	0.8771
29		d/2 = 80.0	0.8747	0.8412
20		d/2 = 90.0	0.8746	0.8077
21		d/2 = 100	0.8669	0.7974
22		d/2 = 110	0.8507	0.7936