Table 1.
A list of features used to train the second RF classifier.
| Index | Description |
|---|---|
| 1–9 | 3 eigenvalues of Hessian matrices on intensity image at σ = 0.5, 1.0, 2.0 mm |
| 10–15 | 1st Gaussian gradient on intensity and NAF probability volumes at σ = 0.36, 0.7, 1.4 mm |
| 16–18 | Intensity, Gaussian smoothed intensity, and NAF probability volumes |
| 19–20 | Laplacian derivative of intensity volume at σ = 0.36, 0.7 mm |
| 21 | Gabor texture feature |
| 22–25 | Intensity statistics: mean, variance, skewness and kurtosis of a 2 mm3 region centered around each graph node |
| 26–28 | Haar features (1.5 mm kernel) along horizontal, vertical & diagonal directions |
| 29–30 | 1D directional gradient along the search column direction on NAF probability and intensity volume |