Research Article

Feature Selection Based on Machine Learning in MRIs for Hippocampal Segmentation

Table 1

The 315 features extracted from the 3D MRI images. Of each group of 66 Haralick features, 13 are the gradients along the 13 diagonals, 5 are the principal moments, and the rest are the three sets of 16 textural features, one set for each plane of the voxels. The gradients for each voxel are measured in all directions at one voxel distance and the relative 3D positions are included as features.

Number Description

1 Position
1 Grey level
66 Haralick features for mask
66 Haralick features for mask
66 Haralick features for mask
66 Haralick features for mask
49 Haar-like 3D features