Ambient Inclusion Trails vs Biological Microboring

Computational framework for discriminating AITs from biological tubular microcavities using morphometric analysis and Fisher LDA

Accuracy: 100.0% 10 Features 60 Samples Noise-Robust

Feature Comparison: AIT vs Biological

Feature Importance (Fisher LDA)

Proxy Combination Accuracy

Robustness Under Noise

Diagnostic Thresholds

FeatureThresholdAccuracyEffect Sizep-value
Branching Index0.0001.000-3.1313.10e-17
Diameter CV0.0230.983-1.8353.09e-09
Straightness0.1590.98338.3303.97e-76
Tortuosity1.0000.983-3.0241.28e-16
Mineral Lining0.4300.9834.2672.95e-23
Organic Residue0.1040.983-3.7867.85e-21
Polygonality0.1890.9672.8621.16e-15
Wall Roughness0.1130.783-0.8651.69e-03
Terminal Shape0.0090.767-1.0561.70e-04
Diameter Trend0.00010.7500.0867.43e-01

Feature Statistics

FeatureAIT MeanAIT StdBio MeanBio Std
Diameter CV0.0210.0010.2020.140
Straightness1.000<0.0010.0820.034
Polygonality0.3530.1210.0970.037
Branching Idx0.0000.0002.8111.270
Wall Roughness0.1010.0080.1270.041
Tortuosity1.000<0.00114.5046.315
Terminal Shape0.0010.0030.1080.143
Mineral Lining0.6920.1380.1580.110
Organic Residue0.0480.0270.5560.188
Diameter Trend0.0002<0.001-0.00020.008