Purpose: A respiratory-phase-specific CBCT scatter correction algorithm, using a 4D virtual CT (4DvCT) as prior, was developed for time-resolved proton therapy dose calculations for lung tumors. The time-resolved dose calculation accuracy achievable with the resulting 4DCBCTcor, as well as with the 4DvCT, was evaluated using a porcine lung phantom allowing reproducible 4DCBCT and 4DCT scanning.
Methods: 4DCBCT projections of a periodically breathing ex-vivo porcine lung phantom with 4 injected artificial lesions, as well as 4DCTs with two different breathing patterns (planning and day-of-treatment), were acquired. The average planning CT was rigidly registered onto the day-of-treatment 3DCBCT. Subsequently, CT mid-position images were computed using deformable image registration (DIR). Exploiting the deformation vector fields, obtained from the previous step, a 4DCBCT was reconstructed using MA-ROOSTER, followed by mid-position image generation. Applying DIR on the two mid-position images yielded a mid-position vCT, which was propagated to 4D with the inverted vector fields from the 4DCBCT mid-positioning. All DIRs were performed using a diffeomorphic Morphons algorithm. DRRs of each of the 4DvCT phases, assumed to be scatter-free, were individually exploited to scatter correct the corresponding CBCT projections. The corrected projection set was again reconstructed with the MA-ROOSTER yielding the 4DCBCTcor.The 4DvCT and 4DCBCTcor methods were evaluated by comparing dose calculations for lesion-specific robust proton plans vs. day-of-treatment reference 4DCT for individual phases and accumulated doses.
Results: Accumulated dose DVH differences of the 4 plans for D(2%), D(98%), D(mean,ITV) and D(mean,lung), were 1.7%, 1.3%, 0.9% and 1.3% for 4DvCT-4DCT and 2.0%, 1.6%, 1.0% and 1.2% for 4DCBCTcor-4DCT. The 2%/2mm gamma pass-rate using a 10% dose threshold was always above 93.2% (4DvCT) and 94.2% (4DCBCTcor).
Conclusion: The qualitative and quantitative agreement between the different modalities suggests that 4DCBCTcor provides sufficient image quality to perform accurate time-resolved proton dose calculation for the used porcine lung phantom.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the German Research Foundation (DFG): project number 399148265, graduate college GRK2274.