Airway tree segmentation for optimal stent placement in image-guided radiotherapy

To reduce the target volume in image-guided radiotherapy it has been proposed to insert a removable thermo-expandable stent near the tumor, enabling the use of motion tracking and advanced gating techniques. An algorithm for extracting geometrical data on the airway tree from CT scans is presented, enabling clinicians to place a stent in close proximity to a lung tumor. The algorithm segments and skeletonizes the airway tree using wavefront propagation. Interpolated virtual slices are placed orthogonal on the skeleton, and the diameter is estimated using 2-D active contour modeling. The segmentation was compared to traditional region growing using seven CT scans of human lungs to validate the performance. The diameter estimation was validated using five phantom CT scans containing known diameters. Validation showed that wavefront propagation in general segmented a larger extent of the airway tree than region growing.