ESTRO 38 Abstract book

S96 ESTRO 38

scan has the highest accuracy in proton range predictions, compared to the Twin-beam DECT method and traditional SECT calibration. OC-0188 Development and commissioning of a set-up optimization routine for ocular proton therapy G. Elisei 1 , R. Via 2 , A. Pella 1 , G. Calvi 4 , R. Ricotti 1 , B. Tagaste 1 , G. Fontana 1 , M.R. Fiore 5 , M. Ciocca 6 , F. Valvo 5 , G. Baroni 1,2 1 Centro Nazionale di Adroterapia Oncologica CNAO, Clinical Department- Bioengineering Unit, Pavia, Italy ; 2 Center for Proton Therapy, Paul Scherrer Institut, Switzerland ; 4 Centro Nazionale di Adroterapia Oncologica CNAO, Particle Accelerator Department, Pavia, Italy ; 5 Centro Nazionale di Adroterapia Oncologica CNAO, Clinical Department, Pavia, Italy ; 6 Centro Nazionale di Adroterapia Oncologica CNAO, Clinical Department- Medical Physics Unit, Pavia, Italy ; 7 Politecnico di Milano, Department of Electronics Information and Bioengineering, Milano, Italy Purpose or Objective Proton therapy is an effective therapeutic option in the treatment of ocular melanomas. Treatment plan is optimized by defining two patients gazing angles (polar and azimuthal) and a brass collimator is used to shape the beam. The patient actively participates in the treatment by fixating a light source, prepositioned according to the treatment plan prescription. At our facility, the fixation light is embedded in an eye tracking (ETS) device which also features optical cameras for eye motion monitoring. The device is mounted on a dedicated robotic arm to guarantee the utmost precision and repeatability in ETS positioning. The system setup is depicted in Figure 1 (a). The close distance between the collimator and the treatment isocenter (7cm) makes for a complicated and sensible treatment setup, particularly considering that the patient wears eyelid retractors and an immobilization mask. Thus, an optimization of the ETS position is required to guarantee a fixation point consistent with the treatment plan while avoiding collisions with collimator and the patient and obstruction of the beam path. In this study we present the development and commissioning of a software application designed to optimize at a patient specific level the ETS position during treatment geometry planning. Material and Methods The ETS position optimizer is built on a three-dimensional rendering of the treatment setup including all elements: fixed (collimator holder), patient-specific (beam path and patient immobilization mask) and moving (ETS). For each gaze direction, the optimizer evaluates more than 600 ETS positions, consistent with the patient’s gazing angles, and select the one that maximizes distances and avoids collisions between all elements of the setup. To evaluate the capability of properly modelling the treatment scenario (Figure 1(b)), we have used the isocentric stereoscopic X-ray imaging system installed in the treatment room. The ETS and the thermoplastic mask were fitted with radiopaque beads and their 3D position, estimated with the optimizer, was compared to the one measured using radiographies. Errors in the optimizer model was quantified in terms of deviation from the planned gazing angles. The analysis was performed on 11 patients treated between Oct 2017 and Jul 2018, using their specific masks and treatment plans.

were first optimized on SECT then forward calculated on the DECT images. The 1D knife-edge prompt gamma camera system was used to measure the prompt gamma signal profile emitted during the proton beam delivery on a spot by spot basis in pencil beam scanning mode. The shifts in proton range were derived by comparing the recorded PGI profiles with simulations on the different CT images. Aggregation with nearby spots (4 and 8 mm radius) was applied to reduce statistical uncertainty of the retrieved shifts. The calculated range shift detected by the PGI camera was first calibrated on a rectangular water phantom to determine the systematic offset of the PGI system which was then applied to the Rando phantom Figure 1 shows the retrieved proton range shifts with a two iso-energy layer proton plan. The median spot shifts for a 4mm spot aggregation were 6.8 mm/2.7 mm for SECT, 3.3 mm/1.6 mm for SQCT, and -3.3 mm/-4.9 mm for TBCT for 190 MeV (range= 23.8 cm water) and 205 MeV (range= 27.1 cm water) protons respectively. studies. Results

Figure 2 compared the retrieved shifts for a prostate plan obtained with simulations on three different CT calibrations. For all the layers within the camera’s field of view (FOV), SQCT calibration yielded the best agreement with measurements compared to SECT or TBCT.

Conclusion Among the different CT calibrations, the sequential DECT