ESTRO 2020 Abstract Book

S898 ESTRO 2020

PO-1639 Surface Image Guided Radiotherapy for breast treatments on Halcyon F. Crop 1 , P. Comte 1 , J. Laffarguette 1 , R. Cayez 1 , D. Pasquier 2 , X. Mirabel 3 , T. Lacornerie 1 1 Centre Oscar Lambret, Medical Physics, Lille, France ; 2 Centre Oscar Lambret & CRIStAL, Academic department of radiation oncology & CRIStAL UMR CNRS 1189, Lille, France ; 3 Centre Oscar Lambret, Academic department of radiation oncology, Lille, France Purpose or Objective The combination of Surface Image Guided Radiotherapy (SIGRT) on halcyon is new. A conventional three camera Catalyst setup around the patient is not possible as the Halcyon system is a closed gantry system. Therefore, three cameras are positioned in front of the gantry. The central camera is closer and oriented towards the real isocenter to enable real-time tracking and breath hold techniques inside the gantry. We installed Catalyst with a slightly different geometry on two halcyon systems (H1 and H2) in order to test the compromise between patient position precision and motion tracking inside the gantry. A Halcyon system requires a daily Quality Control (QC) with a dedicated phantom (MPC). We evaluate if the daily SIGRT QC can be combined with the daily machine QC. Material and Methods We evaluate the precision of the SIGRT breast patient positioning by analyzing the results of Cone Beam CT (CBCT) to planning CT-matching of 1721 sessions. We evaluate the variance differences between H1 and H2 setups with the modified Brown-Forsythe-Levine test (BFL). The precision of the MPC phantom as daily QC device for Catalyst was evaluated by manual displacements in all three and combined directions. Results The H1 Catalyst setup performs slightly better than the H2 setup for positioning in the IEC Y (feet-head) direction (BFL test p<0.001) and in the IEC X (left-right) direction (p=0.02). Most likely this is due to the lateral camera angles and the closer central camera. Mean absolute displacements after CBCT were (X,Y,Z, in mm) (2 ; 2.2 ; 1.8) with standard deviations (2.7, 2.9, 2.4): figures 1 and 2. These results are close to the inter-user fusion variability. There was a remaining table flex of 1.9 mm on H1 and 3.6 mm on H2 for breast patients in the ant-post direction (p < 0.001). Applying the Van Herk principle, a PTV margin of (4.3 ; 4.8 ; 4.7) mm suffices for treatment planning on the H1 setup. In practice, it is recommended to not apply this PTV margin for IMRT and apply a robust optimization scheme for respiration and position uncertainty. Manual displacement tests of the MPC phantom result in a mean error vector of 0.3 mm (largest 0.6 mm). Displacements in the Y direction (Head-Feet) have the largest uncertainty associated in the feet direction: mean error of 0.7 mm on H2 (other directions < 0.5 mm).

factors were evaluated separately for the primary tumour (GTV) and the elective nodal volumes (CTV), with different nodal compartments of the CTV evaluated separately. Margins for several different image guidance strategies were estimated using typical values for (systematic & random) setup uncertainties on bony pelvic structures. Margins were calculated with the intention of delivering 90% of the planned dose to the treatment targets for 90% of patients, as appropriate for neoadjuvant treatment. Based on the findings, a web-based app for margin estimation was developed. Results Table 1 summarises relevant sources of treatment delivery uncertainty, as well as data for systematic (Σ) and random (σ) uncertainties. Multiplicative factors for systematic uncertainties (corresponding to α in the van Herk formula) were set to 2.51 (for GTV) and 2.58 (for CTV, see Nijkamp at al 2012, taking deformations into account). The corresponding factor for random uncertainties (β) was set to 1.28. Penumbra (σ pen ) in soft tissue was assumed to be 3mm. Table 2 summarises appropriate margins for daily imaging and setup on bony anatomy. Note that insufficient data were available to estimate cranio-caudal margins for the CTV. A no-action-level (NAL) strategy, with 3 initial daily images plus weekly imaging, resulted in a 1-2mm increase in margin estimates. Reduced delineation uncertainty decreased margins, but increased impact of daily versus NAL imaging.

Conclusion PTV margins have been estimated for rectal cancer LCRT, based on available published evidence. An interactive app, available at http://tiny.cc/5qsqez , provides PTV margins for different scenarios, including variable values for delineation and setup uncertainty. This work can form the basis for IMRT guidelines, with the addition of clinical expertise and consensus.