Comparative analysis of proficiencies of various textures and geometric features in breast mass classification using k-nearest neighbor

Visual Computing for Industry, Biomedicine, and Art

Table 8 Rank-wise lists of the top 20 most discriminative features selected by four different feature selection techniques

Rank	Relief-F		Pearson correlation coefficient		Neighbourhood component analysis		Term variance
Rank	Index	Feature name	Index	Feature name	Index	Feature name	Index	Feature name
1	F99	FD26	F86	FD13	F1	ASM	F45	Stheta
2	F53	Euler number	F99	FD26	F2	Contrast	F44	Sr
3	F55	Solidity	F87	FD14	F3	Correlation	F51	Convex area
4	F56	Extent	F82	FD9	F4	Sum squares	F52	Filled area
5	F4	Sum squares	F83	FD10	F5	Inverse diff moment	F46	Area
6	F20	Mean	F85	FD12	F6	Sum average	F57	Perimeter
7	F87	FD14	F53	Euler number	F7	Sum variance	F48	Minor axis length
8	F6	Sum average	F92	FD19	F8	Sum entropy	F47	Major axis length
9	F7	Sum variance	F84	FD11	F9	Entropy	F54	Equiv diameter
10	F86	FD13	F95	FD22	F10	Diff variance	F21	Variance
11	F66	ZM3–1	F90	FD17	F11	Diff entropy	F53	Euler number
12	F67	ZM31	F33	SL	F12	Info measure1	F50	Orientation
13	F26	Periodicity	F74	FD1	F13	Info measure2	F7	Sum variance
14	F82	FD9	F97	FD24	F14	Max Corr. Coff	F20	Mean
15	F43	H2	F98	FD25	F15	Homogeneity	F24	Coarseness
16	F21	Variance	F124	FD51	F16	Contrast	F4	Sum squares
17	F22	Skewness	F79	FD6	F17	Mean	F16	Contrast
18	F25	Contrast	F43	H2	F18	Energy	F25	Contrast
19	F8	Sum entropy	F81	FD8	F19	Entropy	F6	Sum average
20	F9	Entropy	F80	FD7	F20	Mean	F23	Kurtosis