ESTRO 2020 Abstract book

S933 ESTRO 2020

INTER FRACTIO N

ML (mm )

AP (mm )

CC (mm )

INTRA FRACTIO N

ML (mm )

AP (mm )

CC (mm )

PO-1617 Feasibility of spirometer-guided single breath-hold kV-CBCTs on Halcyon in lung cancer patients L. Delombaerde 1,2 , T. Depuydt 1,2 , P. Berkovic 1,2 , M. Lambrecht 1,2 1 University Hospital Gasthuisberg, Department of Radiation Oncology, Leuven, Belgium ; 2 KU Leuven, Department of Oncology, Leuven, Belgium Purpose or Objective The fast imaging (and radiation delivery) capabilities of the Halcyon linear accelerator (linac) (Varian Medical Systems) allow for single breath-hold kV-CBCT acquisitions (< 20 sec). The improved image quality, compared to free breathing CBCTs, will potentially allow improved automated contouring in adaptive radiotherapy and radiomics analysis. However, prior to performing breath- hold treatments, the intra fraction tumor position reproducibility has to be quantified. In this study we determined the intra-fraction tumor position reproducibility by acquiring single breath-hold kV-CBCTs in the treatment position at the Halcyon linac using the SDX spirometer (Dyn’R). Material and Methods Five locally advanced lung cancer patients participated in this imaging study (in progress). Prior to the simulation appointment, patients were coached in a training session in which patients were familiarized with the SDX spirometer system and an individualized deep inspiration breath-hold level was determined. During the simulation appointment one additional CT scan was acquired in breath-hold (DIBH-CT), after the free breathing scan with iodine contrast and the 4DCT. Patients were treated in free breathing - our standard of care – on the Halcyon linac. During four fractions (fx 2, 6, 11 and 16) patients received one additional kV-CBCT in breath-hold prior to treatment (PRE-CBCT), and one kV- CBCT in breath-hold after treatment (POST-CBCT). The POST-CBCT was matched to the PRE-CBCT using a 6 degree-of-freedom (DoF) automated rigid registration on the vertebrae. Subsequently, a 3 DoF automated rigid registration was performed on the primary tumor. The difference between the registrations quantified the intra- fraction tumor motion. The same procedure was applied to determine the inter-fraction errors by registering the PRE-CBCT to the DIBH-CT. Results One patient (out of five) was unable to maintain a breath- hold during the coaching session and was excluded from the study. All four remaining patients were able to perform the breath-hold CT and all kV-CBCTs. The CBCT acquisition time was 16 sec and all patients performed every CBCT in a single breath-hold. The improved image quality can be qualitatively assessed in figure 1. Motion blurring is limited resulting in a more clearly defined tumor boundary. The intra-fraction uncertainly was small: median 0.1 mm (range -2.5, 3 mm) in the craniocaudal, median 0.5 mm (range -2.3, 1 mm) in the mediolateral and median 0.7 mm (range -2.1, 2.7 mm) in the anteriorposterior direction. The intra-fraction systematic (Σ) and random (σ) errors are shown in the table.

Σ σ

0.7 0.8 1.6 Σ 1.0 2.1 3.2 σ

0.7 0.7 1.0 0.7 1.1 1.6

Conclusion Single breath-hold kV-CBCTs using a spirometer in lung cancer patients appears feasible and results in good intra- fraction reproducibility and improved CBCT image quality. Such an approach might prove beneficial when considering breath-hold delivery in this patient population or when implementing new adaptive delineation and planning strategies. PO-1618 Markerless Real-Time 3D kV Tracking of Lung Tumors During Free Breathing Stereotactic Radiotherapy K. De Bruin 1 , M. Dahele 1 , H. Mostafavi 2 , B. Slotman 1 , W.F.A.R. Verbakel 1 1 Amsterdam University Medical Centers, Radiation Oncology, Amsterdam, The Netherlands ; 2 Varian Medical Systems, Imaging application, Palo Alto- California, USA Purpose or Objective Real-time tracking during stereotactic lung radiotherapy (SBRT) could confirm the target remains inside the planning target volume (PTV). A gantry-mounted kilo- voltage (kV) imaging system can continuously acquire 2D kV images of the target during volumetric modulated arc therapy (VMAT). Markerless tracking of lung tumors in 2D kV images is considered challenging due to over-projection of internal structures and low contrast with surrounding pixels. Using a life-like moving thorax phantom, we investigated 3D markerless tumor tracking on kV fluoroscopy acquired during free-breathing VMAT lung SBRT. The same method was applied to clinical data. Material and Methods The 3D-printed/molded phantom contains 3 lung tumors (each ~4cm 3 ) in different locations, with different densities. It was moved in 3D in TrueBeam developer mode, using a simulated irregular breathing pattern. Planar kV images were acquired at 7frames/s during 11Gy/fraction 10MV FFF VMAT. 2D reference templates of the tumors+4mm were created for each gantry angle using the inspiration phase of a planning 4DCT. The acquired kV images were matched to templates using normalized cross correlation for determining 2D position projection on the kV panel. Respiratory phase-based triangulation of the 2D matched projection was used to determine the 3 rd dimension of target position. 3D target tracking performed on CBCT projections raw data is presented from 5 patients undergoing free-breathing lung SBRT with a 5mm PTV margin.