Image Classification Using Probabilistic Models that Reflect the Internal Structure of Mixels

Investigation of self-sealing in high-strength and ultra-low-permeability concrete in water using micro-focus X-ray CT

“…Such a voxel is called a "mixel" [21][22][23][24]. Based on this, a maximum likelihood thresholding method considering the effect of mixels [21][22][23] was used to set an appropriate threshold, t. The threshold was determined from the histogram obtained in the boxed area A t in Fig. 8(a).…”

“…If a voxel includes both phases (i.e., precipitated and non-precipitated regions), the CT value, and, accordingly, differences in CT values, of this voxel takes an intermediate CT value between these two phases. Such a voxel is called a "mixel" [21][22][23][24]. Based on this, a maximum likelihood thresholding method considering the effect of mixels [21][22][23] was used to set an appropriate threshold, t. The threshold was determined from the histogram obtained in the boxed area A t in Fig.…”

Estimators of Tissue Proportions from X-Ray CT Images

“…An illustration is given in Figure 4. We see that it is an asymmetric U-shaped distribution, which has the same qualitative shape as the beta distribution proposed by Kitamoto and Takagi (1999). However, their choice, and the uniform distributions used by Santago and Gage (1995) and Laidlaw, Fleischer, and Barr (1998), were chosen empirically, whereas our choice is well founded via a particular point spread function.…”

AdaPT: An adaptive preterm segmentation algorithm for neonatal brain MRI

“…Under the assumption of Gaussian distributed classes on logtransformed data, the initial PV class Gaussian parameters can be approximated by a mixed-tissue distribution in each voxel (Kitamoto and Takagi, 1999), with mean equal to the arithmetic weighted average of its composing class parameters weighted by the determinant of the covariance matrix of each class. Thus, μ Ã j=k ¼ Λ j μ j þ Λ k j jμ k .…”