Program Information
A New Sparse Optimization Scheme for Simultaneous Beam Angle and Fluence Optimization in Radiotherapy Planning
H Liu*, P Dong , L Xing , Stanford Univ School of Medicine, Stanford, CA
Presentations
SU-K-FS2-4 (Sunday, July 30, 2017) 4:00 PM - 6:00 PM Room: Four Seasons 2
Purpose: Sparse minimization-based optimization aims to determine the most promising beam angles by approximating the beam angle optimization (BAO) formulation with efficiently computable convex surrogates. Thus far, only limited success has been achieved by using the approach because of the large inaccuracy incurred in the approximation. The purpose of this work is to develop a robust sparse optimization framework based on the most recent development in group variable selection for reliable BAO.
Methods: We propose the incorporation of the group folded concave penalty (gFCP) as a substitution to the -minimization framework. The new formulation is solved by a variation of an existing gradient method, which directly handles the non-smoothness and non-convexity of gFCP. The performance of the proposed scheme is evaluated by using three IMRT cases: a coplanar prostate case, a coplanar head-and-neck case, and a non-coplanar liver case. The proposed scheme was compared with two state-of-the-art alternative schemes, the L2,1-minimization approach and the gradient norm method (GNM). Both plan quality and the computational efficiency were measured to show the advantages of the proposed framework.
Results: The gFCP-based scheme outperforms both counterpart approaches. In particular, gFCP generates better plan quality than -minimization in all three cases at a comparable computational cost and is always faster than the GNM with competitive, if not better, plan quality.
Conclusion: The proposed gFCP-based scheme provides a promising framework for BAO to shorten the IMRT planning time while maintaining the same, if not improved, level of planning quality.
Contact Email: