Automatic Segmentation of Lung Carcinoma Using 3D Texture Features in 18-FDG PET/CT

Figure 2

The trained DTKNN tree along with the three best features chosen for each node. Note that this DTKNN was trained to discriminate between the tumor ROIs defined by an oncologist and the normal tissue ROIs defined by a medical physicist for a set of 21 patient PET/CT images.