Purpose: Motion management is crucial in scanned proton therapy for mobile tumours. Current motion mitigation approaches rely on single 4DCTs before treatment, ignoring respiratory variability. We investigate the consequences of respiratory variations on internal target volumes (ITV) definition and motion mitigation efficacy, and propose a probabilistic ITV based on 4DMRI.

Materials and methods: Four 4DCT(MRI) datasets, each containing 40 variable cycles of synthetic 4DCTs, were generated by warping single-phase CTs of two lung patients with motion fields extracted from two 4DMRI datasets. Two-field proton treatment plans were optimised on ITVs based on different parts of the 4DCT(MRI)s. 4D dose distributions were calculated by considering variable respiratory patterns. Different probabilistic ITVs were created by incorporating the voxels covered by the CTV in at least 25%, 50%, or 75% (ITV25, ITV50, ITV75) of the cycles, and compared with the conservative ITV encompassing all possible CTV positions.

Results: Depending on the selected planning 4DCT, ITV volumes vary up to 20%, resulting in significant variation in CTV coverage for 4D treatments. Target coverage and homogeneity improved with the conservative ITV, but was associated with significantly increased lung dose (~1%). ITV25 and ITV50 led to acceptable plan quality in most cases without lung dose increments. ITV75 best minimised lung dose, but was insufficient to ensure coverage under all motion scenarios.

Conclusion: Irregular respiration significantly affects CTV coverage when ITVs are only defined by single 4DCTs. A probabilistic ITV50 provides an adequate compromise between target coverage and lung dose for most motion and patient scenarios investigated.