Static couch non-coplanar arc selection optimization for lung SBRT treatment planning

Abstract

Abstract Objective. Non-coplanar arc geometry optimizations that take advantage of beam’s eye view (BEV) geometric overlap information have been proven to reduce dose to healthy organs-at-risk (OARs). Recently, a metric called mean arc distance (MAD) has been developed that quantifies the arc geometry sampling of 4π space. The purpose of this research is to combine improved BEV overlap information with MAD to generate static couch lung stereotactic body radiotherapy (SBRT) treatment plans deliverable on a C-arm linear accelerator. Approach. An algorithm utilizing the Moller–Trumbore ray-triangle intersection method was employed to compute a cost surrogate for dose to overlapping OARs using distances interpolated onto a PDD. Cost was combined with MAD for 100 000 random combinations of arc trajectories. A pathfinding algorithm for arc selection was created, balancing the contributions of MAD and 4π cost for the final trajectory. This methodology was evaluated for 18 lung SBRT patients. Cases were also planned with arcs from a clinical treatment template protocol for dosimetric and plan quality comparison. Results were evaluated using dose constraints in the context of RTOG0915. Main results. Five of six OARs had maximum dose reductions when planned with the arc trajectory optimization algorithm. Significant maximum dose reductions were found for esophagus (7.41 ± 0.91 Gy, p = 0.00019), trachea (5.56 ± 1.55 Gy, p = 0.0025), spinal cord (2.87 ± 1.13 Gy, p = 0.039), large bronchus (3.47 ± 1.49 Gy, p = 0.0075), and aorta (3.13 ± 0.99 Gy, p = 0.012). Mean dose to contralateral lung was also significantly reduced (0.50 ± 0.06 Gy, p = 0.00019). There were two significant increases in OAR doses: mean dose to ipsilateral lung (0.40 ± 0.09, p = 0.00086) and V5Gy to ipsilateral lung (1.95 ± 0.70%, p = 0.011). Paddick conformity index increased by 0.03 ± 0.02 (p = 0.14), remaining below a limit of 1.2 for both techniques. Significance. Static couch non-coplanar optimization yielded maximum dose reductions to OARs while maintaining target conformity for lung SBRT.