Author: Mireille Conrad, Cédric De Marco, Marie Fargier-Voiron, Maud Jaccard, Oscar Matzinger, Anthony Nagy, Stéphane Nevesny, Nicolas Joel Lionel Perichon 👨🔬
Affiliation: Clinique De Genolier 🌍
Purpose: Real-time tracking is implemented on Radixact by detecting the target position with 2D-kV image either directly (for lung tumors), or indirectly (using fiducials as surrogates), and by adapting jaws position for tracking in the longitudinal direction, or MLC opening for vertical or lateral tracking. We aimed to quantify the tracking accuracy depending on the direction and magnitude of target shift.
Methods: Targeting accuracy was quantified by adapting the Cyberknife End-To-End (E2E) methodology. All tracking modes were evaluated: fiducial, fiducial with respiratory, and lung with respiratory. Plans were generated with RayStation12A on three adapted anthropomorphic phantoms, with planned dose distribution centered on two perpendicular radiochromic films. The targeting accuracy (Total Targeting Error TTE) was calculated by comparing the irradiated dose distribution center to the film's geometric center using Accuray’s E2E software. Similar plans were delivered on the Delta4 phantom (fiducial tracking) first centered, then uncentered with gradual shifts of 2 mm in longitudinal and vertical direction. Gamma distribution and dose profiles were compared between centered and shifted plans.
Results: TTE values ranged between 0.42 and 0.71 mm without any tracking, or only jaws tracking. When MLC tracking was needed, values increased to 1.6 mm. Gamma distributions were not affected by longitudinal shifts, whereas MLC correction resulted in periodic shifts of the dose distribution profiles. Gamma passing rates were superior to 90% for shifts near one MLC leaf thickness, but dropped below 85% for shifts near half a leaf thickness, and further decreased with shifts greater than 20 mm.
Conclusion: Our results indicate that Synchrony can deliver plans with minor deviations. The impact of target shifts varies by direction. Jaws corrections are continuous, while MLC corrections are discrete and affected by the FFF profile when large off-axis shifts are corrected. Further investigation is needed to evaluate the impact on treatment plans.