Medical Physics Seminar – Monday, April 29, 2013
Effects of Respiratory Motion on Tumor Subvolume Imaging and Treatment
Andy Ellis (student of Dr. Wolfgang Tome)
Research Assistant, Department of Medical Physics, UW-School of Medicine & Public Health, Madison, WI - USA –
Effects of Respiratory Motion on Risk-Adaptive Radiotherapy Planning
Purpose: Studies have shown that non-uniform dose distribution based on radiosensitivity information obtained through functional imaging can lead in improved tumor response without additional normal tissue side effects. These treatments can be more complex to plan and deliver, and this is compounded in anatomical locations affected by respiration. This research aims to assess imaging artifacts and resulting effects on treatment plans.
Methods and Materials: In order to experimentally analyze the effects caused by non-uniform targets under respiratory motion, a programmable motion lung phantom was constructed. The lung phantom holds various inserts that can be used for imaging or therapy research. For this particular study, 5 PET QA spheres were loaded with non-uniform activity to simulate a tumor with multiple regions of radiosensitivity. The target was driven in 1D S-I motion using patient breathing traces obtained via Varian Real Time Positioning Management (RPM) Respiratory Motion System. 4D and 3D PET/CT imaging was performed for 5 patient breathing patterns scaled to 1, 3, and 5 voxel peak-to-peak motion amplitude. The resulting images were compared using Philips Pinnacle TPS to assess effects on subvolume images resulting from respiratory motion. Non-uniform IMRT plans have been generated to visualize the effects propagation of risk-region misclassification to treatment plans.
Results: Images for a given breathing pattern exhibit expected trends of a positive relationship between 3D PET motion artifacts and motion amplitude. This is noticeably improved when examining individual bins from 4D PET reconstructed images. Autocontoured target subvolumes, however, appear inaccurate spatially. This is confirmed through quantitative comparison of the PET autocontoured subvolumes to the CT manual contoured subvolumes, and is primarily a result of partial volume effects. Non-uniform IMRT plans generated using the PET autocontours as the target underdose the true target by up to 7.2% (low-risk region) and 18.1% (high-risk region).
Conclusions: For non-uniform targets of the size used in this experiment, both respiratory motion and partial volume effects can limit the ability to accurately define subvolumes for image-guided radiotherapy. Comparisons of PET autocontoured treatment subvolumes to “gold-standard†CT manual contoured subvolumes show significant risk-region misclassification, especially for the small high-risk subvolume. Treatment plans generated from both the gold standard contours and PET contours show significant underdose to the true target, with focus again on the small high-risk region.