Search

Introduction SBRT for mobile tumors requires a breathing control method. Since 2014, the dedicated SBRT system VERO (Brainlab-Mitsubishi) provides a real-time dynamic tracking method: “dynamic tracking” (DT). We present this method and the results obtained for quality control, PTV volume and treatment time reduction. Methods The robotic irradiation head of the VERO system (6 MV) is equipped with a 5 mm MLC and follows in real time the target position, thanks to gyroscopic movements guided by infrared and radiological control. DT method was used for liver and pancreas (41 and 3 patients) and thoracic lesions (60 patients) for which the movement amplitude, evaluated by 4DCT (RPM Varian), was >7 mm and after marker implantation (Visicoil, IBA). The treatment plan (TPS IPLAN, Brainlab) was performed with 6 to 8 non-coplanar beams and the Monte Carlo algorithm for lungs on the exhale phase. PTV Margins were determined with the Van Herk et al. 2000 method. PT V DT volume (using DT) was compared to PT V ITV (calculated with MIP images for Internal Target Volume, ITV method) for 60 patients. Treatment times were evaluated for each session for 104 patients. DT mechanical and dosimetric accuracy were evaluated with QUASAR Respiratory Motion et CIRS 008A dynamic thorax phantoms, EBT3 films associated to Doselab Pro software, and with log files analysis. Quality control for 104 patients was performed with Brainlab mobile platform and calibrated films. Results PTV margins are 5 mm for lung and 8 mm for liver. The mean PT V DT was 28.8 cc (6.5 to 14.3 cc) and 46.4 cc for PT V ITV (10.4 to 139 cc), so a 40% volume reduction. The mean treatment time for DT was 25 min, lower than the one for ITV method for which control CBCT are necessary. The patient breathing rates (frequency and amplitude) are often irregular during treatment and different from the reference 4DCT. This doesn’t affect the treatment delivery and target tracking. With 4 year a follow-up, the clinical tolerance during and after SBRT with DT is very good with 1 case of symptomatic grade 3 radiation pneumonitis (PR) and 8 asymptomatic PR for 60 patients. The DT mecanical accuracy was measured γ 1 (3%-1 mm). Conclusion DT allows treatments with breathing control for all patients, even those with irregular breathing rates. Treatment times are lower than classical treatments based on ITV with an accuracy

Introduction The amorphous silicon (a-Si) electronic portal imaging devices (EPIDs) mounted on linacs have been extensively studied and have shown dose-response characteristics suitable for dose verification. Their use for dosimetric purposes is already in clinical routine. However, the response of EPID in continuous acquisition mode was not so studied in literature, and in less extent when considering it for dosimetric applications in stereotactic body radiotherapy (SBRT) conditions (i.e. small fields, and high dose rates using flattening filter free (FFF) beams). The purpose of this work was therefore to characterize the EPID response in terms of relevant dosimetric properties, mainly for the potential use of continuous mode for further SBRT dosimetry with dynamic techniques. Methods The EPID used in this study was the aSi-1000 model installed on a Varian TrueBeamSTx. Its response was characterized in FFF photon beams with nominal energies of 6 and 10 MV and maximum available dose rates (1400 and 2400 MU/min, for 6 and 10 MV respectively). Different sets of images were acquired in both integrated and continuous modes, with EPID at 150 cm from the source, to study dose linearity, repeatability and reproducibility of response, field-size dependence, ghosting effect and response to energy spectrum variations. Before data acquisition, dark field and flood filed images were acquired for offset and gain corrections. As fluctuations in dose rate can potentially influence the response of EPID when using dynamic techniques with continuous acquisition mode, the linearity of EPID response was further evaluated by irradiating a 10 × 10 cm2 field in 20° and 90° arcs focusing on low MU range (3–100 MU), and was also compared to static irradiations. Results Generally, comparable results in terms of dose response linearity, and signal stability for both acquisition modes were obtained, although slightly better for integrated mode. As example, the comparison of EPID response-to-MU ratios for both modes and for the evaluation using the arc field is presented in Fig. 1, for the 6MV FFF. Variations in linearity response were within 2.5% and 2.9%, respectively for the integrated and continuous mode, at lower MUs. Download : Download high-res image (245KB) Download : Download full-size image Conclusions The characterization of the aSi-1000 EPID response to relevant dosimetric properties was investigated, showing the feasibility of its use for dosimetry with FFF beams and high dose rates. Results for continuous mode, which differed from dosimetric non-linearity reported in previous published studies [1] , also encourage the use of EPID for SBRT with dynamic techniques in future works for 3D dosimetry applications (e.g. quality assurance including in vivo dosimetry).

Introduction Since the release of complexes techniques in external radiotherapy, the EPID (Electronic Portal Imaging Device) is more often used to perform patient quality assurance. The purpose of this work was to develop a back-projection algorithm that uses EPID transit images in order to reconstruct the 3D absorbed dose within the patient treated with VMAT plans on the TrueBeam linear accelerator. Materials and Methods Images were acquired with the aS1200 EPID panel embedded on the TrueBeam linear accelerator (Varian) for 10MV beam energy and for a 600 MU/s dose rate. All images were acquired for a 150 cm distance from the source. They were automatically corrected for the dark-field (DF) and for the flood-field (FF) with the dedicated software (Xi, VARIAN). After pre-treating the portal images, a calibration step was performed to transform the grey levels of the pixels into absorbed dose in water via a response function. Correction kernels were also used to correct for the scatter within the EPID. The dose was then back-projected into the patient for all EPID parallel planes for each gantry angle. Finally the total dose was obtained by summing the 3D dose associated for each gantry angulation. The 3D dose reconstructed from the transit EPID images was compared to the one calculated by the TPS (Treatment Planning System) with a global gamma index of 3% and 3 mm criterions. Our algorithm has been tested for 20 VMAT prostate and head and neck treatments. Results The global gamma index obtained is greater than 97% for all prostate treatments and over 96% for all head and neck treatments. Conclusion We have developed and validated a reconstruction algorithm that can be used to reconstruct in vivo the 3D dose distribution in the patient VMAT plans from transit EPID images.

The purpose of our work was to investigate the feasibility of using an EPID-based in-vivo dosimetry method initially designed for conformal fields on pelvic dynamic IMRT fields. The method enables a point dose delivered to the patient to be calculated from the transit signal acquired with an electronic portal imaging device (EPID). After defining a set of correction factors allowing EPID pixel values to be converted into absolute doses, several tests on homogeneous water-equivalent phantoms were performed to estimate the validity of the method in reference conditions. The effects of different treatment parameters, such as delivered dose, field size dependence and patient thickness were also studied. The model was first evaluated on a group of 53 patients treated by 3D conformal radiotherapy (3DCRT) and then on 92 patients treated by IMRT, both for pelvic cancers. For each measurement, the dose was reconstructed at the isocenter (DREC) and compared with the dose calculated by our treatment planning system (DTPS). Excellent agreement was found between DREC and DTPS for both techniques. For 3DCRT treatments, the mean deviation between DREC and DTPS for the 211 in-vivo dose verifications was equal to −1.0 ± 2.2% (1SD). Concerning IMRT treatments, the averaged deviation for the 418 fields verified was equal to −0.3 ± 2.6% (1SD) proving that the method is able to reconstruct a dose for dynamic IMRT pelvic fields. Based on these results, tolerance criteria and action levels were established before its implementation in clinical routine.

Purpose In this work, Artificial Neural Networks (ANN) are used in External Beam Radiotherapy, especially for dosimetry purposes using Electronic Portal Imaging Device (EPID), which needs frequent calibration and complex setting of dedicated software. This work is preliminarily oriented towards quality assurance of intensity-modulated radiotherapy (IMRT). This first step will allow adjusting ANN before their implementation in more complex treatments and in vivodosimetry. Methods Like humans, artificial neural networks work via two phases – learning and recognition. The supervised learning phase entails the creation of “neurons” by linking input and output data. The recognition phase consists of using input data to predict output data. During the learning phase, EPID signals and desired absorbed dose distributions (taken from treatment planning system – TPS) are used as the neural network’s input and output, respectively. Once the learning is complete, new EPID signal can be used to predict the delivered absorbed dose distribution, allowing comparison with the TPS. EPID images were taken during 6MV treatment delivery, with a dose rate of 600 MU/min, from aSi1000 EPID (Varian). Mounted with Exact-arm on a Clinac 23iX equipped with a multi-leaf collimator (120 leaves). The EPID were acquired using the half-resolution mode. 2D plane images were calculated in EclipseTM at the maximum depth dose in a water phantom. Results Learning was performed using 11 input/output datasets from IMRT treatments. All of the used datasets (both EPID inputs and absorbed dose distribution outputs) consisted of 384 × 512 pixels. Learning can be time consuming but once the ANN has been fixed, its use during the recognition phase will be instantaneous. The gamma index, γ, was used to evaluate the difference between the ANN calculated and planned distributions. γ gives the number of pixels (as a percentage) that respect a given objective. γ(2% , 2 mm) for Head and Neck cancers was found to be 99.7%, highlighting the ANN capability to predict the absorbed dose distribution based on EPIDs. Conclusions It was shown that patient-specific quality assurance of IMRT based on EPID can be performed with neural networks algorithms. Next work would be extending algorithms for in vivo dosimetry purpose.

Introduction In order to measure the 2D dose in-vivo from EPID images made on patients, an algorithm that takes into account the heterogeneity has been developed using Matlabtm v.R2015b. Methods The realized algorithm allowed to perform the calculations from the aS-500/2 imager of the clinac 23 iX (Varian) for a 6MV beam with a dose rate of 300UM.min-1. To overcome the lack of uniformity of the detector response and to consider the influence of the backscattered from the robotic arm, a pretreatment was applied to the images. Then the grey levels of the pixels were converted into an absorbed dose to water using correction kernels. Afterward this dose was backprojected in the isocenter plan by applying the inverse square distances law and by taking into account the attenuation of the beams in the medium. To calculate the attenuation factor in each pixel, an additional Matlab module has been developed. First, a Delaunay triangulation was performed to obtain the coordinates points of the entrance and exit on the surface of the patient. Then, the actual distances traveled by each beam in the volume has been determined using a ray-tracing algorithm adapted from the Moller and Trumbore research work. These distances were then converted into water equivalent distances using the Bresenham method. In order to optimize the time calculation, an intermediate step which allowed to segment the patient into boxes was performed. Our algorithm was evaluated on the CIRS thorax phantom and on 20 IMRT beams from different plans optimized for prostate and head & neck treatment. The dose estimated from the EPID was compared to the TPS dose calculation with a γ -index evaluation: 3%–3 mm in the phantom and 5%–3 mm in patients. Results Our optimization method improved the execution time from 30 min to 26 s. On another hand, compared to the algorithm developed in homogeneous medium, this method improved the results in a heterogeneous medium from 80% to 95% of points having a γ -index >1 with the 5%–3 mm criteria. Conclusions The developed dose calculation algorithm can be used to accurately reconstruct in 2D the dose delivered to the patient during treatment with IMRT from EPID images.

Introduction An algorithm to convert EPID signal (SEPID) into dose in water (DWATER) in transit conditions (i.e. with patient) is proposed and evaluated. This work is a necessary first step to perform patient dose reconstruction during his treatment. Materials and methods Conversion of EPID signal into dose is performed according the 4 following steps: 1. Conversion of (SEPID) into relative dose (DREL) thanks to a relative dose response function (fRD) defined for a reference field size (cREF). 2. Deconvolution of DREL with a field size dependance kernel (Kc). 3. Convolution of the array with a dose to water redistribution kernel (Kw). 4. Use of a transmission function to take into account EPID response variation induced by the modification of the beam spectrum energy in case of transit dosimetry. Development and evaluation of our algorithm were perform using in-house MatlabTMsoftware. On a first part, the dose determined from SEPID was compared to ionization chamber measurements on beam axis for several fields. Then, dose profiles were reconstructed from SEPID and compared with those calculated by our treatment planning system for field size ranging from 3 × 3 cm 2 to 20 × 20 cm 2 . Results The relative dose response function established for cREF = 10 × 10 cm 2 was found to be linear. Indeed, fRD can be written as SEPID = a * DWATER + b. Kernels were both modeled as 2D functions dependent on pixel coordinates ( x,y ). O Kc ( x,y ) is an exponential function. O Kw ( x,y ) is a Gaussian function. For all fields, the deviation between the calculated dose and the one measured with the ionization chamber was within ±0.5%. Concerning dose profiles, more than 90% of studied points have a gamma index below than 1 (3%-3 mm). Conclusion All results obtained from our model are in line with experimental measurements. It also confirms that the proposed method will allow a 2D and 3D in vivo dose reconstruction from EPID signal.