Skip to main content
. 2017 May 3;8(28):45888–45897. doi: 10.18632/oncotarget.17585
Autocorrelation = ij(pxμx)(pyμy)σxσy; (1)
Contrast = nn2{ijp(i,j)},|ij|=n; (2)
Correlation = ij(iμx)(jμy)p(i,j)σxσy; (3)
Cluster prominence = ij(i+jμxμy)4p(i,j); (4)
Cluster shading = ij(i+jμxμy)3p(i,j); (5)
Dissimilarity = ijp(i,j)|ij|; (6)
Energy = ijp(i,j)2; (7)
Entropy = ijp(i,j)log(p(i,j)); (8)
Homogeneity = ij11+ijp(i,j); (9)
Difference variance = ii2pxy(i); (10)
Difference entropy = ipx+y(i)log(px+y(i))
HXYHXY1max{HX,HY}
HXY=(8),;
(11)
Information measure of correlation = HXY1=ijp(i,j)log(px(i)py(j))
HX=entropy of px,
HY=entropy of py;
(12)
Inverse difference normalized = ij11+|ij|p(i,j); (13)
Inverse difference moment = ij11+(ij)2p(i,j); (14)