The authors have demonstrated that motion information from 4D-MRI can be used to generate realistic 4D-CT data sets on the basis of a single static 3D-CT data set. 4D-CT(MRI) presents a novel approach to test the robustness of treatment plans in the circumstance of patient motion.

Advanced 4D dose calculations (4DDCs) for scanned particle therapy show that in the incidence of motion, it is insufficient to use target contours defined on one reference CT phase. ICRU Report 62 (ICRU 1999 ICRU Report 62 (Bethesda, MD: ICRU)) advises that variations in size, shape and position of CTVs relative to anatomic reference points have to be considered for internal target volumes (ITVs). In addition to geometrical margin adaption, changes of water equivalent path length have to be considered for particle therapy. Different ITV concepts have been applied to six representative patients (liver and lung indications) based on 4DCT. Geometrical ITVs (gITV) were calculated by combining deformed CTVs over all motion phases. To take into account path length changes, range adapted ITVs (raITV) were established as the union of range adapted CTVs in all phases. For gated delivery, gat_gITVs and gat_raITVs were calculated. Extensive 4DDCs have been performed for two exemplary patients to illustrate that neither re-scanning nor gating can sufficiently compensate for motion effects if no appropriate margins are employed and to evaluate the effectiveness of gITVs and raITVs. CTVs significantly differ from gITVs and raITVs in size (up to a factor 2 in volume). But also raITVs and gITVs differ significantly in size and are spatially displaced, particularly for lung patients. raITVs show a strong field dependence in shape. All volumes are reduced in size when gating is applied and considered during margin adaption. 4D dose distributions show big improvements when gITV or raITV are used compared to CTVs. However, the use of either gITVs or raITVs do not result in significant differences. If raITVs are used, slightly better target coverage is gained at the cost of more healthy tissue exposure. Our results emphasize that adapted target volumes have to be used for scanned particle therapy in the presence of motion. However, even though gITVs and raITVs differ significantly in shape and size, this difference does not necessarily translate into significant differences in the resultant 4D dose distributions.

Proton therapy with active scanning beam delivery has significant advantages compared to conventional radiotherapy. However, so far only static targets have been treated in this way, since moving targets potentially lead to interplay effects. For 4D treatment planning, information on the target motion is needed to calculate time-resolved dose distributions. In this study, respiratory liver motion has been extracted from 4D CT data using two deformable image registration algorithms. In moderately moving patient cases (mean motion range around 6 mm), the registration error was no more than 3 mm, while it reached 7 mm for larger motions (range around 13 mm). The obtained deformation fields have then been used to calculate different time-resolved 4D treatment plans. Averaged over both motion estimations, interplay effects can increase the D₅-D₉₅ value for the clinical target volume (CTV) from 8.8% in a static plan to 23.4% when motion is considered. It has also been found that the different deformable registration algorithms can provide different motion estimations despite performing similarly for the selected landmarks, which in turn can lead to differing 4D dose distributions. Especially for single-field treatments where no motion mitigation is used, a maximum (mean) dose difference (averaged over three cases) of 32.8% (2.9%) can be observed. However, this registration ambiguity-induced uncertainty can be reduced if rescanning is applied or if the treatment plan consists of multiple fields, where the maximum (mean) difference can decrease to 15.2% (0.57%). Our results indicate the necessity to interpret 4D dose distributions for scanned proton therapy with some caution or with error bars to reflect the uncertainties resulting from the motion estimation. On the other hand, rescanning has been found to be an appropriate motion mitigation technique and, furthermore, has been shown to be a robust approach to also deal with these motion estimation uncertainties.