Dosimetric evaluation of respiratory gating on a 0.35-T magnetic resonance-guided radiotherapy linac.

Abstract

The commercial 0.35-T magnetic resonance imaging (MRI)-guided radiotherapy vendor ViewRay recently introduced upgraded real-time imaging frame rates based on compressed sensing techniques. Furthermore, additional motion tracking algorithms were made available. Compressed sensing allows for increased image frame rates but may compromise image quality. To assess the impact of this upgrade on respiratory gating accuracy, we evaluated gated dose distributions pre- and post-upgrade using a motion phantom and radiochromic film. Seven motion waveforms (four artificial, two patient-derived free-breathing, and one breath-holding) were used to drive an MRI-compatible motion phantom. A treatment plan was developed to deliver a 3-cm diameter spherical dose distribution typical of a stereotactic body radiotherapy plan. Gating was performed using 4-frames per second (fps) imaging pre-upgrade on the "default" tracking algorithm and 8-fps post-upgrade using the "small mobile targets" (SMT) and "large deforming targets" (LDT) tracking algorithms. Radiochromic film was placed in a moving insert within the phantom to measure dose. The planned and delivered dose distributions were compared using the gamma index with 3%/3-mm criteria. Dose-area histograms were produced to calculate the dose to 95% (D95) of the sphere planning target volume (PTV) and two simulated gross tumor volumes formed by contracting the PTV by 3 and 5mm, respectively. Gamma pass rates ranged from 18% to 93% over the 21 combinations of breathing trace and gating conditions examined. D95 ranged from 206 to 514cGy. On average, the LDT algorithm yielded lower gamma and D95 values than the default and SMT algorithms. Respiratory gating at 8fps with the new tracking algorithms provides similar gating performance to the original algorithm with 4fps, although the LDT algorithm had lower accuracy for our non-deformable target. This indicates that the choice of deformable image registration algorithm should be chosen deliberately based on whether the target is rigid or deforming.