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

Table 1

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

estimation			MRI Alzheimer’s	Skin lesion	MRI brain
1	Classical mean		0.8841	0.8703	0.8406
2	Lognormal [6]		0.8796	0.8860	0.8479
3	Harmonic		0.8902	0.8697	0.8440
4	Geometric		0.8722	0.8376	0.8513
5	Contraharmonic	Q = −3.0	0.8893	0.8446	0.8549
6		Q = −1.0	0.8943	0.8697	0.8440
7		Q = −0.5	0.8971	0.8669	0.8512
8		Q = 0.0	0.8703	0.8702	0.8407
9		Q = 0.1	0.8887	0.8603	0.8449
10	Alpha trim	d/2 = 10.0	0.8924	0.8714	0.8667
11		d/2 = 20.0	0.8971	0.8727	0.8830
12		d/2 = 30.0	0.8987	0.8739	0.8942
13		d/2 = 40.0	0.8973	0.8795	0.8899
14		d/2 = 50.0	0.9046	0.8876	0.8779
15		d/2 = 55.0	0.8968	0.8917	0.8789
16		d/2 = 60.0	0.8922	0.8824	0.8777
17		d/2 = 65.0	0.8846	0.8721	0.8724
18		d/2 = 70.0	0.8928	0.8679	0.8782
29		d/2 = 80.0	0.8842	0.8692	0.8645
20		d/2 = 90.0	0.8831	0.8689	0.8633
21		d/2 = 100	0.8787	0.8632	0.8536
22		d/2 = 110	0.8755	0.8581	0.8534