Room: Exhibit Hall | Forum 5
Purpose: To study the uncertainties in 4D CT planning of lung SBRT tumors using dose accumulation based in deformable image registration (DIR).
Methods: Three patients were planned and treated of lung SBRT. Planning was done based on 4D CT images according to the ITV concept, using 3D CRT and IMRT with 6 and 10 MV FFF photon beams from a Elekta Versa linear accelerator. Dose calculation was performed in the reference phase with Pinnacle v9.10. Patients were treated with a compression belt, to limit respiratory motion, which was in all cases below 15 mm of amplitude. To assess the uncertainties due to motion and deformation, the dose distributions were retrospectively calculated in all the 4D CT phases. MIM software was used to perform a DIR and dose accumulation on the reference phase. Dose indicators were calculated on the reference phase and accumulated dose matrices for the GTV and the PTV (D95 and Dmin, calculated as D0.3cm3, the minimum dose to the tumor not taking into consideration the 0.3 cm3 with minimum dose values).
Results: Dose accumulation revealed differences in GTV coverage according to D95 and D0.3cm3 to be small (see Table). Differences are greater in the PTV, leading up to -8.6% for the D95 and -5.3% for the D0.3cm3 in one case (dose coverage was worse according to full 4D dose calculation and accumulation). In the three cases, the degradation in the dose coverage for both metrics in the PTV was always greater than -4%.
Conclusion: Dose accumulation based on DIR revealed dose differences in the coverage of the PTV. This means that due to treatment planning uncertainties assumed with the ITV approach there is a probability of GTV underdosage. More data are needed to assess the clinical impact of this finding.
Funding Support, Disclosures, and Conflict of Interest: The authors acknowledge the financial support from the Spanish Ministerio de Economia y Competitividad (Instituto de Salud Carlos III through FIS project PI16/00899) and the European Regional Development Fund (ERDF).