Validation and testing of a novel pencil-beam model derived from Monte Carlo simulations in carbon-ion treatment planning for different scenarios

Deep learning-based fast denoising of Monte Carlo dose calculation in carbon ion radiotherapy

BACKGROUND

Plan verification is one of the important steps of quality assurance (QA) in carbon ion radiotherapy. Conventional methods of plan verification are based on phantom measurement, which is labor-intensive and time-consuming. Although the plan verification method based on Monte Carlo (MC) simulation provides a more accurate modeling of the physics, it is also time-consuming when simulating with a large number of particles. Therefore, how to ensure the accuracy of simulation results while reducing simulation time is the current difficulty and focus.

PURPOSE

The purpose of this work was to evaluate the feasibility of using deep learning-based MC denoising method to accelerate carbon-ion radiotherapy plan verification.

METHODS

Three models, including CycleGAN, 3DUNet and GhostUNet with Ghost module, were used to denoise the 1 × 106 carbon ions-based MC dose distribution to the accuracy of 1 × 108 carbon ions-based dose distribution. The CycleGAN's generator, 3DUNet and GhostUNet were all derived from the 3DUNet network. A total of 59 cases including 29 patients with head-and-neck cancers and 30 patients with lung cancers were collected, and 48 cases were randomly selected as the training set of the CycleGAN network and six cases as the test set. For the 3DUNet and GhostUNet models, the numbers of training set, validation set, and test set were 47, 6, and 6, respectively. Finally, the three models were evaluated qualitatively and quantitatively using RMSE and three-dimensional gamma analysis (3 mm, 3%).

RESULTS

The three end-to-end trained models could be used for denoising the 1 × 106 carbon ions-based dose distribution, and their generalization was proved. The GhostUNet obtained the lowest RMSE value of 0.075, indicating the smallest difference between its denoised and 1 × 108 carbon ions-based dose distributions. The average gamma passing rate (GPR) between the GhostUNet denoising-based versus 1 × 108 carbon ions-based dose distributions was 99.1%, higher than that of the CycleGAN at 94.3% and the 3DUNet at 96.2%. Among the three models, the GhostUNet model had the fewest parameters (4.27 million) and the shortest training time (99 s per epoch) but achieved the best denoising results.

CONCLUSION

The end-to-end deep network GhostUNet outperforms the CycleGAN, 3DUNet models in denoising MC dose distributions for carbon ion radiotherapy. The network requires less than 5 s to denoise a sample of MC simulation with few particles to obtain a qualitative and quantitative result comparable to the dose distribution simulated by MC with relatively large number particles, offering a significant reduction in computation time.

Feasibility study of hybrid inverse planning with transmission beams and single-energy spread-out Bragg peaks for proton FLASH radiotherapy

BACKGROUND

Ultra-high dose rate (FLASH) proton planning with only transmission beams (TBs) has limitations in normal tissue sparing. The single-energy spread-out Bragg peaks (SESOBPs) of the FLASH dose rate have been demonstrated feasible for proton FLASH planning.

PURPOSE

To investigate the feasibility of combining TBs and SESOBPs for proton FLASH treatment.

METHODS

A hybrid inverse optimization method was developed to combine the TBs and SESOBPs (TB-SESOBP) for FLASH planning. The SESOBPs were generated field-by-field from spreading out the BPs by pre-designed general bar ridge filters (RFs) and placed at the central target by range shifters (RSs) to obtain a uniform dose within the target. The SESOBPs and TBs were fully placed field-by-field allowing automatic spot selection and weighting in the optimization process. A spot reduction strategy was conducted in the optimization process to push up the minimum MU/spot assuring the plan deliverability at beam current of 165 nA. The TB-SESOBP plans were validated in comparison with the TB only (TB-only) plans and the plans with the combination of TBs and BPs (TB-BP plans) regarding 3D dose and dose rate (dose-averaged dose rate) distributions for five lung cases. The FLASH dose rate coverage (V40Gy/s ) was evaluated in the structure volume receiving > 10% of the prescription dose.

RESULTS

Compared to the TB-only plans, the mean spinal cord D1.2cc drastically reduced by 41% (P < 0.05), the mean lung V7Gy and V7.4 Gy moderately reduced by up to 17% (P < 0.05), and the target dose homogeneity slightly increased in the TB-SESOBP plans. Comparable dose homogeneity was achieved in both TB-SESOBP and TB-BP plans. Besides, prominent improvements were achieved in lung sparing for the cases of relatively large targets by the TB-SESOBP plans compared to the TB-BP plans. The targets and the skin were fully covered with the FLASH dose rate in all three plans. For the OARs, V40Gy/s  = 100% was achieved by the TB-only plans while V40Gy/s  > 85% was obtained by the other two plans.

CONCLUSION

We have demonstrated that the hybrid TB-SESOBP planning was feasible to achieve FLASH dose rate for proton therapy. With pre-designed general bar RFs, the hybrid TB-SESOBP planning could be implemented for proton adaptive FLASH radiotherapy. As an alternative FLASH planning approach to TB-only planning, the hybrid TB-SESOBP planning has great potential in dosimetrically improving OAR sparing while maintaining high target dose homogeneity.

A probability model for anatomical robust optimisation in head and neck cancer proton therapy

Objective.Develop an anatomical model based on the statistics of the population data and evaluate the model for anatomical robust optimisation in head and neck cancer proton therapy.Approach.Deformable image registration was used to build the probability model (PM) that captured the major deformation from patient population data and quantified the probability of each deformation. A cohort of 20 nasopharynx patients was included in this retrospective study. Each patient had a planning CT and 6 weekly CTs during radiotherapy. We applied the model to 5 test patients. Each test patient used the remaining 19 training patients to build the PM and estimate the likelihood of a certain anatomical deformation to happen. For each test patient, a spot scanning proton plan was created. The PM was evaluated using proton spot location deviation and dose distribution.Main results. Using the proton spot range, the PM can simulate small non-rigid variations in the first treatment week within 0.21 ± 0.13 mm. For overall anatomical uncertainty prediction, the PM can reduce anatomical uncertainty from 4.47 ± 1.23 mm (no model) to 1.49 ± 1.08 mm at week 6. The 95% confidence interval (CI) of dose metric variations caused by actual anatomical deformations in the first week is -0.59% ∼ -0.31% for low-risk CTD₉₅, and 0.84-3.04 Gy for parotidD_mean. On the other hand, the 95% CI of dose metric variations simulated by the PM at the first week is -0.52 ∼ -0.34% for low-risk CTVD₉₅, and 0.58 ∼ 2.22 Gy for parotidD_mean.Significance.The PM improves the estimation accuracy of anatomical uncertainty compared to the previous models and does not depend on the acquisition of the weekly CTs during the treatment. We also provided a solution to quantify the probability of an anatomical deformation. The potential of the model for anatomical robust optimisation is discussed.