Self-Trained Supervised Segmentation of Subcortical Brain Structures Using Multispectral Magnetic Resonance Images

Figure 2

The randomly generated trainings points (in red) overlapped to a T1w axial slice for one of the studies of the Image dataset I (IXI002-Guys-0828). Spatial coordinates and the corresponding multispectral intensities of these points were used by the classifiers for the learning phase.