Program Information
Aperture Control for VMAT Delivery Systems
W Henao1*, M Epelman1 , M Matuszak1 , E Romeijn2, K Younge1, C Anderson1,(1) University of Michigan, Ann Arbor, Michigan, (2) Georgia Institute of Technology, Atlanta, GA
Presentations
MO-F-CAMPUS-TT-2 (Monday, July 31, 2017) 4:30 PM - 5:30 PM Room: Therapy ePoster Theater
Purpose: VMAT is an efficient and highly utilized planning and delivery technique. However, VMAT plan optimization, especially for complex cases, can result in highly irregular aperture shapes and lead to inefficient and inaccurate deliveries. Previous attempts to control VMAT complexity during optimization require inefficient stochastic optimization techniques such as simulated annealing. Here, we introduce a framework to efficiently penalize VMAT aperture complexity during optimization using column generation methods.
Methods: To make VMAT complexity penalization feasible in a clinical timeframe, we have extended a column generation model-based VMAT optimization framework to support an aperture shape objective. The model iteratively adds a new control point and its corresponding aperture shape and intensity at each iteration. An aperture complexity penalization term, which is monotone increasing in the area of the aperture and monotone decreasing in the perimeter, is added to the conventional dose based objectives with a user- selected weight. The algorithm performance was validated using a head/neck case. Increasing penalty was placed on aperture complexity to observe the relationship between plan quality and aperture complexity.
Results: The addition of the aperture complexity penalty to the column generation based VMAT algorithm adds less than twice the time to the conventional optimization, which is considered to be clinically feasible in practice. Aperture complexity was reduced with increasing penalty, as expected. A relative weighting factor of 10 reduced aperture complexity by 19.1% while at the same time increasing the conventional objective function cost by 1.49%. Further automation in choosing the penalty weight may be possible in different disease sites.
Conclusion: Complexity reduction in VMAT aperture was achieved by means of an efficient column generation VMAT optimization algorithm. In our head/neck example, aperture complexity can be reduced without significant effect on dose volume metrics and increasing algorithm runtime by 69.72%.
Funding Support, Disclosures, and Conflict of Interest: Martha Matuszak has a research grant from Varian Medical Systems, unrelated to the current work. Kelly Young is listed in the same grant. This project is funded in part by NIH P01-CA059872
Contact Email: