International Journal of Molecular Imaging / 2013 / Article / Fig 9

Clinical Study

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

Figure 9

Results from the voxel-by-voxel DTKNN classification (red) shown with the oncologist's GTV contour (green) overlaid on transverse slices of a fused PET/CT volumes (top left, top right, and bottom right). The highest performing PET threshold of 30% max SUV (purple) is also shown overtop of the best performing hard class defined by the 3-FLAB algorithm (teal).