Table 2 First and second order textural markers.
From: A novel computer-aided diagnostic system for accurate detection and grading of liver tumors
Textural marker | Definition |
|---|---|
First order | |
Mean (\(\mathrm {\mu }\)) | Represents the gray-level values balance point of each object. It is calculated simply by getting the average gray-level value for each object. |
Variance | Describes the gray-level distribution around our computed Mean. |
Skewness | Expresses how the gray-level values are asymmetrically distributed around the Mean of the object. |
Kurtosis | Measures to what extent the gray-level values are concentrated towards the tails of the distribution. |
Entropy | Expresses the amount of randomness within each structure gray-level values. |
CDFs | Return the cumulative distribution function of the histogram density values. This is calculated along the whole object and getting the cumulative sum of the gray-level values (Normalized to [0 to 1] at multiple positions (from 0 to 100% of the object with a 10% step). |
Percentiles | Calculate the percentiles of gray-level values for the corresponding CDFs. |
Second order | |
Contrast | Measures the disparity in gray-level values between neighbors. |
Dissimilarity | Finds to what extent voxels are different from their neighbors. |
Homogeneity | Expresses the inverse difference moment among neighbors. |
Angular second moment (ASM) | Determines the gray-levels local uniformity (orderliness). |
Energy | The square root of the ASM. |
Correlation | Determines the gray-level linear dependency between center voxel and its neighbors. |
Gray-level non-uniformity (GLN) | Describes the dissimilarity of gray-level values within the object. |
High gray-level run emphasis (HGLRE) | Measures the concentration of high gray-level values in the structure. |
Long run emphasis (LRE) | Determines how long run lengths are distributed in the object indicating the coarseness of the texture. |
Long run high gray-level emphasis (LRHGLE) | Measures how long runs of high gray-level values are distributed in the object. |
Long run low gray-level emphasis (LRLGLE) | Measures how long runs of low gray-level values are distributed in the object. |
Low gray-level run emphasis (LGLRE) | Measures the concentration of low gray-level values in the structure. |
Run entropy (RE) | Indicates the amount of randomness in gray-level runs in the structure. |
Run length non-uniformity (RLN) | Expresses the inhomogeneity among run lengths in the object. |
Run percentage (RP) | Is calculated by the division of the overall count of runs by the total number of pixels. |
Short run emphasis (SRE) | Measures the concentration of short run lengths in the object indicating how fine the texture is. |
Short run high gray-level emphasis (SRHGLE) | Measures the concentration of high gray-level values short runs in the object. |
Short run low gray-level emphasis (SRLGLE) | Measures the concentration of low gray-level values short runs in the object. |
