Texture Discretization Sensitivity

Sweep GLCM bin width on a synthetic textured lesion and watch contrast, energy and entropy shift by large factors even though the tissue never changes.

  • GLCM
  • discretization
  • IBSI
  • texture
  • gray-level binning
Gray levels used
18.000
Contrast
4.701
Energy
0.016
Entropy
4.420

Discretized phantom

18 gray levels

Continuous phantom (reference)

native intensity units

Feature value vs. bin width

05101520253001020304050Bin width (units)Feature value
ContrastEntropyEnergy ×20

Sensitivity across the bin-width sweep

Coefficient of variation across the sweep — contrast: 1.64, energy: 1.10, entropy: 0.53.
Bin widthGray levelsContrastEnergyEntropyHomogeneity
1.00064.00048.8620.0026.7830.252
2.00044.00023.3980.0036.1400.315
3.00030.00011.1650.0065.3910.389
5.00018.0004.7010.0164.4200.493
8.00012.0002.2870.0393.5190.598
12.0009.0001.2540.0802.8120.692
16.0007.0000.9370.1302.3120.750
20.0006.0000.9640.1931.9650.790
25.0005.0000.6960.2711.6660.829
32.0004.0000.8790.3031.4450.837
Texture features are not properties of the tumor alone
The lesion never changes across this sweep — only the discretization does. Contrast and entropy shift by large factors as bin width changes because they are computed on the quantized image, not the physical intensity field. IBSI-style reporting requires fixing the bin width (or the number of gray levels) and stating it explicitly; comparing "contrast" values computed at different bin widths compares different quantities that happen to share a name.