No abstract

Purpose: Quality assurance (QA) of complex linear accelerators is critical and highly time consuming. Varian’s Machine Performance Check (MPC) uses IsoCal phantom to test geometric and dosimetric aspects of the TrueBeam systems in <5min. In this study we independently tested the accuracy and robustness of the MPC tools. Methods: MPC is automated for simultaneous image‐acquisition, using kV‐and‐MV onboard‐imagers (EPIDs), while delivering kV‐and‐MV beams in a set routine of varying gantry, collimator and couch angles. MPC software‐tools analyze the images to test: i) beam‐output and uniformity, ii) positional accuracy of isocenter, EPIDs, collimating jaws (CJs), MLC leaves and couch and iii) rotational accuracy of gantry, collimator and couch. 6MV‐beam dose‐output and uniformity were tested using ionization‐chamber (IC) and ICarray. Winston‐Lutz‐Tests (WLT) were performed to measure isocenter‐offsets caused by gantry, collimator and couch rotations. Positional accuracy of EPIDs was evaluated using radio‐opaque markers of the IsoCal phantom. Furthermore, to test the robustness of the MPC tools we purposefully miscalibrated a non‐clinical TrueBeam by introducing errors in beam‐output, energy, symmetry, gantry angle, couch translations, CJs and MLC leaves positions. Results: 6MV‐output and uniformity were within ±0.6% for most measurements with a maximum deviation of ±1.0%. Average isocenter‐offset caused by gantry and collimator rotations was 0.316±0.011mm agreeing with IsoLock (0.274mm) and WLT (0.41mm). Average rotation‐induced couch‐shift from MPC was 0.378±0.032mm agreeing with WLT (0.35mm). MV‐and‐kV imager‐offsets measured by MPC were within ±0.15mm. MPC predicted all machine miscalibrations within acceptable clinical tolerance. MPC detected the output miscalibrations within ±0.61% while the MLC and couch positions were within ±0.06mm and ±0.14mm, respectively. Gantry angle miscalibrations were detected within ±0.1°. Conclusions: MPC is a useful tool for QA of TrueBeam systems and its automation makes it highly efficient for testing both geometric and dosimetric aspects of the machine. This is very important for hypo‐fractionated SBRT treatments. Received support from Varian Medical Systems, Palo Alto, CA 94304‐1038.

Purpose: Proton treatment planning systems are not able to accurately predict output factors and do not calculate monitor units (MU) for proton fields. Output factors (cGy/MU) for patient‐specific fields are usually measured in phantoms or modeled empirically. The purpose of this study is to predict the output factors (OFs) for a given proton (R90) and modulation width (Mod) for the first Mevion S250 proton therapy system. Methods: Using water phantoms and a calibrated ionization chamber‐electrometer, over 100 OFs were measured for various R90 and Mod combinations for 24 different options. OFs were measured at the center of the Mod, which coincided with the isocenter. The measured OFs were fitted using an analytic model developed by Kooy (Phys.Med.Biol. 50, 2005) for each option and a derived universal empirical‐based polynomial as a function of R90 and Mod for all options. Options are devised for ranges of R90 and Mod. The predicted OFs from both models were compared to measurements. Results: Using the empirical‐based model, the values could be predicted to within 3% for at least 90% of measurements and within 5% for 98% of the measurements. Using the analytic model to fit each option with the same effective source position, the prediction is much more accurate. The maximal uncertainty between measured and predicted is within 2% and the averaged root‐mean‐square is 1.5%. Conclusion: Although the measured data was not exhaustive, both models predicted OFs within acceptable uncertainty. Both models are currently used for a sanity check of our continual patient field OF measurements. As we acquire more patient‐field OFs, the model will be refined with an ultimate goal of eliminating the time‐consuming patient‐specific OF measurements.

Purpose: I‐Beam, a 3D‐ultrasound guided prostate localization system, uses Coordinate Positioning System (CPS) Technology in registering live ultrasound images of the patient to the treatment machine coordinate system. The goal of this study is to verify the accuracy of I‐Beam's camera system and its precision in localizing the US images to the isocenter. Method and Materials: The first part of the study was to examine the I‐Beam camera system for integrity in tracking the probe position while translating the US probe in 3D‐space. We then examined alignment correctness when registering the captured US images and contours. A robotic arm of the Nomos BAT system was used for accurate translations. Both the arm and camera system were fixed to the treatment couch. Longitudinal translations of the camera system were confirmed by the Nomos Auto‐crane. An ultrasound phantom mimicking pelvic anatomy was employed in quantifying the precision of the image localization and contour alignment. The phantom was translated to known distances, recapturing the 3D‐ultrasound volume, and aligned the contours to determine the shifts computed by the I‐Beam system. Results: X‐, Y‐ and Z‐ translations recorded by the robotic system were subtracted from the coordinates of the camera system in computing the errors. When the camera system was repositioned within ±10 cm of the 3D‐space, the CPS technology predicted the ultrasound probe position within ±0.5 mm. The absolute deviations were (mean ± SD): 0.3±0.2, 0.2±0.2 and 0.2±0.2 in lateral, longitudinal and vertical directions, respectively. I‐Beam's image registration and contour alignment were accurate within ±1mm. Conclusion: The I‐Beam camera system is accurate to within ± 0.5mm in tracking the US probe position. Phantom measurement showed the US image acquisition and the contour alignment is also accurate within ±1mm. These new tests assured that the I‐Beam system is accurate in tracking, image‐registration and alignment.

Purpose: Creation of a simple robust tool to use in conjunction with onboard fluoroscopic imaging of the patient in treatment position that will allow daily review of target motion relative to the expected motion used to plan treatment. This tool is needed to circumvent software incompatibilities in a multi‐vendor environment. Method and Materials: The procedure creates a radio‐opaque outline of the treatment aperture and physically attaches it the kV detector surface so as to project the aperture on subseuent fluoroscopic images. The patients need to have a treatment target that is visible on fluoroscopy or have an implanted radiopaque fiducial marker. A DRR is printed showing the planned tumor or fiducial range of motion as a projected structure outline. The outline will be used to evaluate the absolute position of the tumor or marker relative to the intended location. The motion range is outlined with a radiopaque wire, and 5 wire crosses are placed for alignment and scale verification. The wired DRR is attached to the kV detector and the patient is imaged with fluoroscopy; displaying the expected motion range (wire outline) against the projected tumor or fiducial marker motion. The wire outline position and scale are checked against a virtual reticule. Results: Using the methods described above it is possible to visualize a moving target just prior to treatment as well as the treatment location. Care must be taken to ensure not only accurate placement of the outline but also that the correct structure is outlined for the current patient. Conclusion: The procedure outlined here creates a flexible process that is non machine, software or manufacturer dependant. The procedure is sensitive to user error and is dependant on the manufacturing and placement accuracy of the device.