An experimental validation of a filtering approach for prompt gamma prediction in a research proton treatment planning system.

Objective.Prompt gamma (PG) radiation generated from nuclear reactions between protons and tissue nuclei can be employed for range verification in proton therapy. A typical clinical workflow for PG range verification compares the detected PG profile with a predicted one. Recently, a novel analytical PG prediction algorithm based on the so-called filtering formalism has been proposed and implemented in a research version of RayStation (RaySearch Laboratories AB), which is a widely adopted treatment planning system. This work validates the performance of the filtering PG prediction approach.Approach.The said algorithm is validated against experimental data and benchmarked with another well-established PG prediction algorithm implemented in a MATLAB-based software REGGUI. Furthermore, a new workflow based on several PG profile quality criteria and analytical methods is proposed for data selection. The workflow also calculates sensitivity and specificity information, which can help practitioners to decide on irradiation course interruption during treatment and monitor spot selection at the treatment planning stage. With the proposed workflow, the comparison can be performed on a limited number of selected high-quality irradiation spots without neighbouring-spot aggregation.Main results.The mean shifts between the experimental data and the predicted PG detection (PGD) profiles (ΔPGD) by the two algorithms are estimated to be1.5±2.1mm and-0.6±2.2mm for the filtering and REGGUI prediction methods, respectively. The ΔPGD difference between two algorithms is observed to be consistent with the beam model difference within uncertainty. However, the filtering approach requires a much shorter computation time compared to the REGGUI approach.Significance.The novel filtering approach is successfully validated against experimental data and another widely used PG prediction algorithm. The workflow designed in this work selects spots with high-quality PGD shift calculation results, and performs sensitivity and specificity analyses to assist clinical decisions.

An updated variable RBE model for proton therapy.

Objective.While a constant relative biological effectiveness (RBE) of 1.1 forms the basis for clinical proton therapy, variable RBE models are increasingly being used in plan evaluation. However, there is substantial variation across RBE models, and several newin vitrodatasets have not yet been included in the existing models. In this study, an updatedin vitroproton RBE database was collected and used to examine current RBE model assumptions, and to propose an up-to-date RBE model as a tool for evaluating RBE effects in clinical settings.Approach.A proton database (471 data points) was collected from the literature, almost twice the size of the previously largest model database. Each data point included linear-quadratic model parameters and linear energy transfer (LET). Statistical analyses were performed to test the validity of commonly applied assumptions of phenomenological RBE models, and new model functions were proposed forRBEmaxandRBEmin(RBE at the lower and upper dose limits). Previously published models were refitted to the database and compared to the new model in terms of model performance and RBE estimates.Main results.The statistical analysis indicated that the intercept of theRBEmaxfunction should be a free fitting parameter and RBE estimates were clearly higher for models with free intercept.RBEminincreased with increasing LET, while a dependency ofRBEminon the reference radiation fractionation sensitivity (α/ßx) did not significantly improve model performance. Evaluating the models, the new model gave overall lowest RMSE and highest R2 score. RBE estimates in the distal part of a spread-out-Bragg-peak in water (α/ßx= 2.1 Gy) were 1.24-1.51 for original models, 1.25-1.49 for refits and 1.42 for the new model.Significance.An updated RBE model based on the currently largest database among published phenomenological models was proposed. Overall, the new model showed better performance compared to refitted published RBE models.

Proton range uncertainty caused by synthetic computed tomography generated with deep learning from pelvic magnetic resonance imaging.

BACKGROUND: In proton therapy, it is disputed whether synthetic computed tomography (sCT), derived from magnetic resonance imaging (MRI), permits accurate dose calculations. On the one hand, an MRI-only workflow could eliminate errors caused by, e.g., MRI-CT registration. On the other hand, the extra error would be induced due to an sCT generation model. This work investigated the systematic and random model error induced by sCT generation of a widely discussed deep learning model, pix2pix. MATERIAL AND METHODS: An open-source image dataset of 19 patients with cancer in the pelvis was employed and split into 10, 5, and 4 for training, testing, and validation of the model, respectively. Proton pencil beams (200 MeV) were simulated on the real CT and generated sCT using the tool for particle simulation (TOPAS). Monte Carlo (MC) dropout was used for error estimation (50 random sCT samples). Systematic and random model errors were investigated for sCT generation and dose calculation on sCT. RESULTS: For sCT generation, random model error near the edge of the body (∼200 HU) was higher than that within the body (∼100 HU near the bone edge and <10 HU in soft tissue). The mean absolute error (MAE) was 49 ± 5, 191 ± 23, and 503 ± 70 HU for the whole body, bone, and air in the patient, respectively. Random model errors of the proton range were small (<0.2 mm) for all spots and evenly distributed throughout the proton fields. Systematic errors of the proton range were -1.0(±2.2) mm and 0.4(±0.9)%, respectively, and were unevenly distributed within the proton fields. For 4.5% of the spots, large errors (>5 mm) were found, which may relate to MRI-CT mismatch due to, e.g., registration, MRI distortion anatomical changes, etc. CONCLUSION: The sCT model was shown to be robust, i.e., had a low random model error. However, further investigation to reduce and even predict and manage systematic error is still needed for future MRI-only proton therapy.

Data-driven ion-independent relative biological effectiveness modeling using the beam quality Q.

Beam quality Q = Z2/E (Z = ion charge, E = energy), an alternative to the conventionally used linear energy transfer (LET), enables ion-independent modeling of the relative biological effectiveness (RBE) of ions. Therefore, the Q concept, i.e. different ions with similar Q have similar RBE values, could help to transfer clinical RBE knowledge from better-studied ion types (e.g. carbon) to other ions. However, the validity of the Q concept has so far only been demonstrated for low LET values. In this work, the Q concept was explored in a broad LET range, including the so-called overkilling region. The particle irradiation data ensemble (PIDE) was used as experimentalin vitrodataset. Data-driven models, i.e. neural network (NN) models with low complexity, were built to predict RBE values for H, He, C and Ne ions at differentin vitroendpoints taking different combinations of clinically available candidate inputs: LET, Q and linear-quadratic photon parameterαx/ßx. Models were compared in terms of prediction power and ion dependence. The optimal model was compared to published model data using the local effect model (LEM IV). The NN models performed best for the prediction of RBE at reference photon doses between 2 and 4 Gy or RBE near 10% cell survival, using onlyαx/ßxand Q instead of LET as input. The Q model was not significantly ion dependent (p > 0.5) and its prediction power was comparable to that of LEM IV. In conclusion, the validity of the Q concept was demonstrated in a clinically relevant LET range including overkilling. A data-driven Q model was proposed and observed to have an RBE prediction power comparable to a mechanistic model regardless of particle type. The Q concept provides the possibility of reducing RBE uncertainty in treatment planning for protons and ions in the future by transferring clinical RBE knowledge between ions.

An ion-independent phenomenological relative biological effectiveness (RBE) model for proton therapy.

BACKGROUND: A relative biological effectiveness (RBE) of 1.1 is used for proton therapy though clinical evidence of varying RBE was raised. Clinical studies on RBE variability have been conducted for decades for carbon radiation, which could advance the understanding of the clinical proton RBE given an ion-independent RBE model. In this work, such a model, linear and simple, using the beam quantity Q = Z2/E (Z = ion charge, E = kinetic energy per nucleon) was tested and compared to the commonly used, proton-specific and linear energy transfer (LET) based Wedenberg RBE model. MATERIAL AND METHODS: The Wedenberg and Q models, both predicting RBEmax and RBEmin (i.e., RBE at vanishing and very high dose, respectively), are compared in terms of ion-dependence and prediction power. An experimental in-vitro data ensemble covering 115 publications for various ions was used as dataset. RESULTS: The model parameter of the Q model was observed to be similar for different ions (in contrast to LET). The Q model was trained without any prior knowledge of proton data. For proton RBE, the differences between experimental data and corresponding predictions of the Wedenberg or the Q model were highly comparable. CONCLUSIONS: A simple linear RBE model using Q instead of LET was proposed and tested to be able to predict proton RBE using model parameter trained based on only RBE data of other particles in a clinical proton energy range for a large in-vitro dataset. Adding (pre)clinical knowledge from carbon ion therapy may, therefore, reduce the dominating biological uncertainty in proton RBE modelling. This would translate in reduced RBE related uncertainty in proton therapy treatment planning.

Accounting for prompt gamma emission and detection for range verification in proton therapy treatment planning.

Prompt gamma (PG) imaging is widely investigated as one of the most promising methods for proton range verification in proton therapy. The performance of this technique is affected by several factors like tissue heterogeneity, number of protons in the considered pencil beam and the detection device. Our previous work proposed a new treatment planning concept which boosts the number of protons of a few PG monitoring-friendly pencil beams (PBs), selected on the basis of two proposed indicators quantifying the conformity between the dose and PG at the emission level, above the desired detectability threshold. To further explore this method at the detection level, in this work we investigated the response of a knife-edge slit PG camera which was deployed in the first clinical application of PG to proton therapy monitoring. The REGistration Graphical User Interface (REGGUI) is employed to simulate the PG emission, PG detection as well as the corresponding dose distribution. As the PG signal detected by this kind of PG camera is sensitive to the relative position of the camera and PG signal falloff, we optimized our PB selection method for this camera by introducing a new camera position indicator identifying whether the expected falloff of the PG signal is centered in the field of view of the camera or not. Our camera-adapted PB selection method is investigated using computed tomography (CT) scans at two different treatment time points of a head and neck, and a prostate cancer patient under scenarios considering different statistics level. The results show that a precision of 0.8 mm for PG falloff identification can be achieved when a PB has more than 2 × 108 primary protons. Except for one case due to unpredictable and comparably large anatomical changes, the PG signals of most of the PBs recommended by all our indicators are observed to be reliable for proton range verification with deviations between the inter-fractional shift of proton range (as deduced from the PB dose distribution) and the detected PG signal within 2.0 mm. In contrast, a shift difference up to 9.6 mm has been observed for the rejected PBs. The magnitude of the proton range shift due to the inter-fractional anatomical changes is observed to be up to 23 mm. The proposed indicators are shown to be valuable for identifying and recommending reliable PBs to create new PG monitoring-friendly TPs. Comparison between our PB boosting method and the alternative PB aggregation, which combines the signal of nearby PBs to reach the desired counting statistics, is also discussed.

A new treatment planning approach accounting for prompt gamma range verification and interfractional anatomical changes.

Prompt gamma (PG) imaging is widely investigated for spot-by-spot in vivo range verification for proton therapy. Previous studies pointed out that the accuracy of prompt gamma imaging is affected by the statistics (number of protons delivered per pencil beam) of the proton beams and the conformity between prompt gamma and dose distribution (PG-dose correlation). Recently a novel approach to re-optimize conventional treatment plans by boosting a few pencil beams with good PG-dose correlation above the statistics limit for reliable PG detectability was proposed. However, up to now, only PG-dose correlation on the planning computed tomography (CT) was considered, not accounting for the fact that the robustness of the PG-dose correlation is not guaranteed in the cases of interfractional anatomical changes. In this work, this approach is further explored with respect to the robustness of the PG-dose correlation of each pencil beam in the case of interfractional anatomical changes. A research computational platform, combining Monte Carlo pre-calculated pencil beams with the analytical Matlab-based treatment planning system (TPS) CERR, is used for treatment planning. Geant4 is used for realistic simulation of the dose delivery and PG generation for all individual pencil beams in the heterogeneous patient anatomy using multiple CT images for representative patient cases (in this work, CTs of one prostate and one head and neck cancer patient are used). First, a Monte Carlo treatment plan is created using CERR. Thereby the PG emission and dose distribution for each individual spot is obtained. Second, PG-dose correlation is quantified using the originally proposed approach as well as a new indicator, which accounts for the sensitivity of individual spots to heterogeneities in the 3D dose distribution. This is accomplished by using a 2D distal surface (dose surface) derived from the 3D dose distribution for each spot. A few pencil beams are selected for each treatment field, based on their PG-dose correlation and dose surface, and then boosted in the new re-optimized treatment plan. All treatment plans are then fully re-calculated with Monte Carlo on the CT scans of the corresponding patient at three different time points. The result shows that all treatment plans are comparable in terms of dose distribution and dose averaged LET distributions. The spots recommended by our indicators maintain good PG-dose correlation in the cases of interfractional anatomical changes, thus ensuring that the proton range shift due to anatomical changes can be monitored. Compared to another proposed spots aggregation approach, our approach shows advantages in terms of the detectability and reliability of PG, especially in presence of heterogeneities.

Toward a new treatment planning approach accounting for in vivo proton range verification.

Protons with modern pencil-beam scanning delivery are widely used in state-of-the-art radiotherapy. To reduce the unwanted effect of proton range uncertainties, prompt gamma (PG) monitoring is investigated and considered one of the most promising methods for real-time, in vivo range verification. Despite good correlation between the penetration depth of the PG signal and proton range in most cases, mismatch can occur especially because of tissue heterogeneities. Moreover, detectability and reproducibility of the prompt gamma signal critically depends on counting statistics. Nowadays, conventional treatment planning systems do not account for the degree of correlation between dose and PG signal nor the expected PG signal counting statistics, which considerably influences the possibility of a reliable verification of the intended beam range. Hence, in this project, we investigate a new treatment planning approach, in which the spot-by-spot conformities between PG and dose profiles (PG-dose correlation) as well as PG signal detectability and precision are taken into account based on a TPS optimizer. To investigate the feasibility of this idea, a research computational platform, combining Monte Carlo (MC, Geant4) pre-calculated pencil beams with the analytical Matlab-based TPS engine CERR, is used for treatment planning. Geant4 is employed for realistic simulation of the dose delivery and PG generation of all spots in the heterogeneous patient anatomy given by CT images. First of all, a treatment plan is created using a charged particle extension of CERR. Secondly, the PG fall-off positions of all individual pencil beams are evaluated and compared to the 80% distal dose fall-off positions. Thirdly, the PG-dose correlations of all spots are quantified. A new plan, in which a few spots with the best PG-dose correlation are boosted to ensure PG detectability with good precision, is then made. Finally, the optimized plan is fully recalculated on the same patient CT using Geant4, and the result is evaluated considering both plan quality and beam range monitorability. The evaluation shows that the re-optimized treatment plan is comparable to the initial plan in terms of dose distribution, dose averaged LET distribution and robustness, while fulfilling the set statistical conditions for reliable PG monitoring of the few automatically or manually selected spots. The method could thus complement, and for the selected pencil beams even overcome limitations of, alternative suggested approach such as pencil beam aggregation to provide sufficient counting statistics for precise PG range retrieval with good correlation to the treatment dose.

Theoretical substantiation of biological efficacy enhancement for ß-delayed particle decay (9)C beam: A Monte Carlo study in combination with analysis with the local effect model approach.

PURPOSE: To improve the efficacy of heavy ion therapy, ß-delayed particle decay (9)C beam as a double irradiation source for cancer therapy has been proposed. The authors' previous experiment showed that relative biological effectiveness (RBE) values at the depths around the Bragg peak of a (9)C beam were enhanced and compared to its stable counterpart (12)C beam. The purpose of this study was to explore the nature of the biological efficacy enhancement theoretically. METHODS: A Monte Carlo simulation study was conducted in this study. First a simplified cell model was established so as to form a tumor tissue. Subsequently, the tumor tissue was imported into the Monte Carlo simulation software package gate and then the tumor cells were virtually irradiated with comparable (9)C and (12)C beams, respectively, in the simulations. The transportation and particle deposition data of the (9)C and (12)C beams, derived from the gate simulations, were analyzed with the authors' local effect model implementation so as to deduce cell survival fractions. RESULTS: The particles emitted from the decay process of deposited (9)C particles around a cell nucleus increased the dose delivered to the nucleus and elicited clustered damages around the secondary particles' trajectories. Therefore, compared to the (12)C beam, the RBE value of the (9)C beam increased at the depths around their Bragg peaks. CONCLUSIONS: Collectively, the increased local doses and clustered damages due to the decayed particles emitted from deposited (9)C particles led to the RBE enhancement in contrast with the (12)C beam. Thus, the enhanced RBE effect of a (9)C beam for a simplified tumor model was shown theoretically in this study.