The purpose of this work is to evaluate the impact of single-, dual- and multi-energy CT (SECT, DECT and MECT) on proton range uncertainties in a patient like geometry and a full Monte Carlo environment. A virtual patient is generated from a real patient pelvis CT scan, where known mass densities and elemental compositions are overwritten in each voxel. Simulated CT images for SECT, DECT and MECT are generated for two limiting cases: (1) theoretical and idealistic CT numbers only affected by Gaussian noise (case A, the best scenario) and (2) reconstructed polyenergetic sinograms containing beam hardening, projection-based Poisson noise, and reconstruction artifacts (case B, the worst scenario). Conversion of the simulated SECT images into Monte Carlo inputs is done following the stoichiometric calibration method. For DECT and MECT, the Bayesian eigentissue decomposition method of Lalonde (2017 Med. Phys. 44 5293-302) is used. Pencil beams from seven different angles around the virtual patient are simulated using TOPAS to assess the performance of each method. Percentage depth doses curves (PDD) are compared to ground truth in order to determine the accuracy of range prediction of each imaging modality. For the idealistic images of case A, MECT and DECT slightly outperforms SECT. Root mean square (RMS) errors or 0.78 mm, 0.49 mm and 0.42 mm on R mm, are observed for SECT, DECT and MECT respectively. In case B, PDD calculated in the MECT derived Monte Carlo inputs generally shows the best agreement with ground truth in both shape and position, with RMS errors of 2.03 mm, 1.38 mm and 0.86 mm for SECT, DECT and MECT respectively. Overall, the Bayesian eigentissue decomposition used with DECT systematically predicts proton ranges more accurately than the gold standard SECT-based approach. When CT numbers are severely affected by imaging artifacts, MECT with four energy bins becomes more reliable than both DECT and SECT.

The purpose of this study is to evaluate the impact of a novel tissue characterization method using dual-energy over single-energy computed tomography (DECT and SECT) on Monte Carlo (MC) dose calculations for low-dose rate (LDR) prostate brachytherapy performed in a patient like geometry. A virtual patient geometry is created using contours from a real patient pelvis CT scan, where known elemental compositions and varying densities are overwritten in each voxel. A second phantom is made with additional calcifications. Both phantoms are the ground truth with which all results are compared. Simulated CT images are generated from them using attenuation coefficients taken from the XCOM database with a 100 kVp spectrum for SECT and 80 and 140Sn kVp for DECT. Tissue segmentation for Monte Carlo dose calculation is made using a stoichiometric calibration method for the simulated SECT images. For the DECT images, Bayesian eigentissue decomposition is used. A LDR prostate brachytherapy plan is defined with I sources and then calculated using the EGSnrc user-code Brachydose for each case. Dose distributions and dose-volume histograms (DVH) are compared to ground truth to assess the accuracy of tissue segmentation. For noiseless images, DECT-based tissue segmentation outperforms the SECT procedure with a root mean square error (RMS) on relative errors on dose distributions respectively of 2.39% versus 7.77%, and provides DVHs closest to the reference DVHs for all tissues. For a medium level of CT noise, Bayesian eigentissue decomposition still performs better on the overall dose calculation as the RMS error is found to be of 7.83% compared to 9.15% for SECT. Both methods give a similar DVH for the prostate while the DECT segmentation remains more accurate for organs at risk and in presence of calcifications, with less than 5% of RMS errors within the calcifications versus up to 154% for SECT. In a patient-like geometry, DECT-based tissue segmentation provides dose distributions with the highest accuracy and the least bias compared to SECT. When imaging noise is considered, benefits of DECT are noticeable if important calcifications are found within the prostate.

Purpose. Respiration-induced motion introduces significant positioning uncertainties in radiotherapy treatments for thoracic sites. Accounting for this motion is a non-trivial task commonly addressed with surrogate-based strategies and latency compensating techniques. This study investigates the potential of a new unified probabilistic framework to predict both future target motion in real-time from a surrogate signal and associated uncertainty. Method. A Bayesian approach is developed, based on a Kalman filter theory adapted specifically for surrogate measurements. Breathing motions are collected simultaneously from a lung target, two external surrogates (abdominal and thoracic markers) and an internal surrogate (liver structure) for 9 volunteers during 4 min, in which severe breathing changes occur to assess the robustness of the method. A comparison with an artificial non-linear neural network (NN) is performed, although no confidence interval prediction is provided. A static worst-case scenario and a simple static design are investigated. Results. Although the NN can reduce the prediction errors from thoracic surrogate in some cases, the Bayesian framework outperforms in most cases the NN when using the other surrogates: bias on predictions is reduced by 38% and 16% on average when using respectively the liver and the abdomen for the simple scenario, and by respectively 40% and 31% for the worst-case scenario. The standard deviation of residuals is reduced on average by up to 42%. The Bayesian method is also found to be more robust to increasing latencies. The thoracic marker appears to be less reliable to predict the target position, while the liver shows to be a better surrogate. A statistical test confirms the significance of both observations. Conclusion. The proposed framework predicts both the future target position and the associated uncertainty, which can be valuably used to further assist motion management decisions. Further investigation is required to improve the predictions by using an adaptive version of the proposed framework.

Background Target motion in the course of radiotherapy is one of the largest factors affecting the treatment quality of highly dynamic sites such as lung. A critical component of real‐time motion management is not only the prediction of tumor location at a future point in time but assessment of positional uncertainty for the purposes of margin adjustment and optimization of validation schemes. Purpose In this study, we propose to investigate the ability of a confidence estimator to accurately reflect the reliability of individual target position predictions and prospectively detect large prediction errors by relying exclusively on a surrogate signal. Methods This work uses a Bayesian framework for indirect tracking. While constant covariance estimates are commonly used to express the uncertainty of the models involved, in this study new adaptive estimates are derived from the surrogate behavior to reflect increasing uncertainty when the breathing conditions differ from the reference conditions observed during the training step. The accuracy of the resulting 95% predicted confidence regions (CRs) is evaluated on nine breathing sequences involving changes of respiratory types (free, thoracic, abdominal, deep). The breathing motions are collected simultaneously from a lung target and two different surrogate signals (an external marker and an anatomical location within the liver). Receiver operating characteristic (ROC) analysis is performed to evaluate the ability of the predicted uncertainty to prospectively detect large prediction errors. Results Higher CR accuracy is obtained when using the proposed adaptive estimates over using constant estimations: on average over the cohort, the proportion of actual target positions lying within the 95% CR is increased by 40 and 35 p.p. with the internal and external surrogates. The time‐dependent inflation of the CR width tends to match the magnitude variation of the prediction errors: the adaptive CR effectively enlarges when the target position cannot be predicted reliably, which corresponds to potentially high prediction errors. More precisely, the ROC analysis indicates that the proposed uncertainty estimate can detect if prediction errors are greater than 5 mm with on average high sensitivity (90%) and modest specificity (54% and 47% from internal and external surrogates, respectively). Conclusions While relying exclusively on the surrogate motion characteristics being continuously monitored, the Bayesian framework coupled to adaptive uncertainty estimations can provide reliable CR able to detect large prediction errors. The findings of this study could be further used to automatically trigger risk management mechanisms prospectively.