ESTRO 38 Abstract book

S1074 ESTRO 38

Baroni 2,6 , M. Ciocca 1 1 National Center for Oncological Hadrontherapy CNAO, Medical Physics Department, Pavia, Italy ; 2 Politecnico di Milano, Department of Electronics- Information and Bioengineering, Milan, Italy ; 3 Heidelberg Ion Beam Therapy Center HIT, Physics Department, Heidelberg, Germany ; 4 National Center for Oncological Hadrontherapy CNAO, Radiology Department, Pavia, Italy ; 5 National Center for Oncological Hadrontherapy CNAO, Radiotherapy Department, Pavia, Italy ; 6 National Center for Oncological Hadrontherapy CNAO, Bioengineering Department, Pavia, Italy Purpose or Objective Irradiation geometry for abdominal targets in which a carbon-ion beam passes at the lung-diaphragm interface for a gated scanning treatment is not robust in our experience of plan recalculations on re-evaluative 4DCTs. In this work the potentialities of patient-specific virtual CTs, deformed on extracted 4DMRI vector fields (VFs) to quantify and mitigate the impact of motion pattern variations in respiratory-gated carbon ion treatments of the abdominal site were investigated. Material and Methods 3 patients (2 liver, 1 pancreas) treated at CNAO with a respiratory-gated carbon-ion plan were selected. In addition to the 4DCT imaging workflow, 4DMRI images were obtained by retrospective sorting of multi-slice 2DMRI acquisition, covering a limited field-of-view of 10 cm in the L-R direction. Virtual CTs were generated by applying the 4DMRI-extracted motion fields (VFs) to the acquired 0%Ex (full-exhale) CT. Range of motion (ROM) for the right kidney was calculated as centroid distance between 100%In (full-inhale) and 0%Ex phases, and the difference was computed between 4DMRI and 4DCT. GTV ROMs were also compared to a population of 10 patients (averaged ROM derived on 4DCT data) treated at our facility. The GTV was then warped from 0%Ex to the other breathing phases in the 4DMRI datasets and the ITV was defined. For each patient, a single-beam carbon-ion plan was optimized delivering uniform dose to the ITV on the 0%Ex CT. Recalculations were carried out both on the 100%In CT and on virtual-CTs depicting 0%Ex and 100%In phases. D 2% , D 50% , D 98% were calculated for ITV and GTV 100In .

Conclusion Consistency between 4DMRI and 4DCT was verified on kidney MR, ITV definition, and by plan recalculation on virtual CTs. Validation on a larger group of patients is needed. EP-1969 Dosimetric effect of diaphragm motion on target volume coverage for oesophageal cancer R. Stansbridge 1 , M. Borland 1 , B. Stewart Thomson 2 , M. Gilmore 1 1 Clatterbridge Cancer Centre, Physics, Liverpool, United Kingdom ; 2 Clatterbridge Cancer Centre, Radiation Services, Liverpool, United Kingdom Purpose or Objective Intra-fraction diaphragm motion for oesophageal cancers can result in dosimetric uncertainty due to changes in the electron density of individual voxels during the breathing cycle. The objective of this study is to quantify the dosimetric impact of diaphragm motion on PTV coverage over the breathing cycle to provide evidence for 3D planned VMAT. Material and Methods Radical oesophagus patients with significant diaphragm overlap of the PTV were identified to demonstrate the largest potential effect on plan dosimetry. All had received a prescription of either 50.4Gy/28# (n=2), 50Gy/25# (n=4) or 55Gy/20# (n=4). At our centre, for patients with a regular breathing trace, a 4DCT is acquired and a VMAT plan produced on the Average Intensity Projection (AIP) with a PTV margin of 0.5cm. For patients who do not have a regular breathing trace, a 3DCT scan is acquired as this may catch the diaphragm at an extreme that is not representative of the average diaphragm position during treatment, these patients are currently treated with a conformal plan. In order to verify the resilience of VMAT plans to this potential discrepancy, individual bins from the 4DCT were used as a surrogate for a 3DCT taken at the extremes of diaphragm motion (peak inspiration and peak expiration). A simulated plan was optimised on each of the two extreme cases with the PTV copied from the original plan. After optimisation, the plan was then recalculated on the AIP to represent the full respiratory motion during treatment delivery and assessed for changes in dosimetry. Results The mean peak to peak diaphragm motion was 2.3 ± 0.9cm with a maximum of 3.5cm. The mean difference from peak inspiration to AIP for D98% was -0.81 ± 0.45% and peak expiration to AIP was 0.48 ± 0.22%. All hotspots remained within the PTV as defined on the AIP with the mean change in D2% from peak inspiration to AIP was -0.26%. Hotspots were enhanced in the plans recalculated from peak expiration to AIP with an average increase of 1.83%. Doses to proximal OAR were also assessed, with minimal changes to mean dose and <0.3Gy change over the whole prescription. It should be noted, however, that structures were not changed from those outlined on the original AIP. Hotspots became apparent in the peak expiration plan onto AIP but still deemed clinically acceptable according to local protocols. The hotspots were the result of change in density from liver to lung tissue and created at the lung- tissue interface within the PTV, the maximum D2% was 106.9% (3.9% change between peak to AIP). Conclusion This study suggests that VMAT plans are sufficiently robust to diaphragm motion to allow VMAT planning of 3D scanned oesophagus patients. Patients will now be treated with an increased PTV margin of 1cm to account for PTV motion not captured. It is also recommended that diaphragm motion is monitored at treatment to ensure consistency. Further investigation is planned for patients with diaphragm motion >4 cm.

Results ROM differences between 4DCT and 4DMRI for the kidney were <2% on average. GTV ROM were within the population range (2-12mm). ITV volume differences were <5%. Target DVH parameters discrepancies between dose distributions evaluated on acquired and virtual CT at corresponding phases were <1% on average.